case study on application of ai

• Digital Marketing
• Facebook Marketing
• Instagram Marketing
• Ecommerce Marketing
• Content Marketing
• Data Science Certification
• Machine Learning
• Artificial Intelligence
• Data Analytics
• Graphic Design
• Adobe Illustrator
• Web Designing
• UX UI Design
• Interior Design
• Front End Development
• Back End Development Courses
• Business Analytics
• Entrepreneurship
• Supply Chain
• Financial Modeling
• Corporate Finance
• Project Finance
• Harvard University
• Stanford University
• Yale University
• Princeton University
• Duke University
• UC Berkeley
• Harvard University Executive Programs
• MIT Executive Programs
• Stanford University Executive Programs
• Oxford University Executive Programs
• Cambridge University Executive Programs
• Yale University Executive Programs
• Kellog Executive Programs
• CMU Executive Programs
• 45000+ Free Courses
• Free Certification Courses
• Free DigitalDefynd Certificate
• Free Harvard University Courses
• Free MIT Courses
• Free Excel Courses
• Free Google Courses
• Free Finance Courses
• Free Coding Courses
• Free Digital Marketing Courses

20 Detailed Artificial Intelligence Case Studies [2024]

In this dynamic era of technological advancements, Artificial Intelligence (AI) emerges as a pivotal force, reshaping the way industries operate and charting new courses for business innovation. This article presents an in-depth exploration of 20 diverse and compelling AI case studies from across the globe. Each case study offers a deep dive into the challenges faced by companies, the AI-driven solutions implemented, their substantial impacts, and the valuable lessons learned. From healthcare and finance to transportation and retail, these stories highlight AI’s transformative power in solving complex problems, optimizing processes, and driving growth, offering insightful glimpses into the potential and versatility of AI in shaping our world.

Related: How to Become an AI Thought Leader?

1. IBM Watson Health: Revolutionizing Patient Care with AI

Task/Conflict: The healthcare industry faces challenges in handling vast amounts of patient data, accurately diagnosing diseases, and creating effective treatment plans. IBM Watson Health aimed to address these issues by harnessing AI to process and analyze complex medical information, thus improving the accuracy and efficiency of patient care.

Solution: Utilizing the cognitive computing capabilities of IBM Watson, this solution involves analyzing large volumes of medical records, research papers, and clinical trial data. The system uses natural language processing to understand and process medical jargon, making sense of unstructured data to aid medical professionals in diagnosing and treating patients.

Overall Impact:

• Enhanced accuracy in patient diagnosis and treatment recommendations.
• Significant improvement in personalized healthcare services.

Key Learnings:

• AI can complement medical professionals’ expertise, leading to better healthcare outcomes.
• The integration of AI in healthcare can lead to significant advancements in personalized medicine.

2. Google DeepMind’s AlphaFold: Unraveling the Mysteries of Protein Folding

Task/Conflict: The scientific community has long grappled with the protein folding problem – understanding how a protein’s amino acid sequence determines its 3D structure. Solving this problem is crucial for drug discovery and understanding diseases at a molecular level, yet it remained a formidable challenge due to the complexity of biological structures.

Solution: AlphaFold, developed by Google DeepMind, is an AI model trained on vast datasets of known protein structures. It assesses the distances and angles between amino acids to predict how a protein folds, outperforming existing methods in terms of speed and accuracy. This breakthrough represents a major advancement in computational biology.

• Significant acceleration in drug discovery and disease understanding.
• Set a new benchmark for computational methods in biology.
• AI’s predictive power can solve complex biological problems.
• The application of AI in scientific research can lead to groundbreaking discoveries.

3. Amazon: Transforming Supply Chain Management through AI

Task/Conflict: Managing a global supply chain involves complex challenges like predicting product demand, optimizing inventory levels, and streamlining logistics. Amazon faced the task of efficiently managing its massive inventory while minimizing costs and meeting customer demands promptly.

Solution: Amazon employs sophisticated AI algorithms for predictive inventory management, which forecast product demand based on various factors like buying trends, seasonality, and market changes. This system allows for real-time adjustments, adapting swiftly to changing market dynamics.

• Reduced operational costs through efficient inventory management.
• Improved customer satisfaction with timely deliveries and availability.
• AI can significantly enhance supply chain efficiency and responsiveness.
• Predictive analytics in inventory management leads to reduced waste and cost savings.

4. Tesla’s Autonomous Vehicles: Driving the Future of Transportation

Task/Conflict: The development of autonomous vehicles represents a major technological and safety challenge. Tesla aimed to create self-driving cars that are not only reliable and safe but also capable of navigating complex traffic conditions without human intervention.

Solution: Tesla’s solution involves advanced AI and machine learning algorithms that process data from various sensors and cameras to understand and navigate the driving environment. Continuous learning from real-world driving data allows the system to improve over time, making autonomous driving safer and more efficient.

• Leadership in the autonomous vehicle sector, enhancing road safety.
• Continuous improvements in self-driving technology through AI-driven data analysis.
• Continuous data analysis is key to advancing autonomous driving technologies.
• AI can significantly improve road safety and driving efficiency.

Related: High-Paying AI Career Options

5. Zara: Fashioning the Future with AI in Retail

Task/Conflict: In the fast-paced fashion industry, predicting trends and managing inventory efficiently are critical for success. Zara faced the challenge of quickly adapting to changing fashion trends while avoiding overstock and meeting consumer demand.

Solution: Zara employs AI algorithms to analyze fashion trends, customer preferences, and sales data. The AI system also assists in managing inventory, ensuring that popular items are restocked promptly and that stores are not overburdened with unsold products. This approach optimizes both production and distribution.

• Increased sales and profitability through optimized inventory.
• Enhanced customer satisfaction by aligning products with current trends.
• AI can accurately predict consumer behavior and trends.
• Effective inventory management through AI can significantly impact business success.

6. Netflix: Personalizing Entertainment with AI

Task/Conflict: In the competitive streaming industry, providing a personalized user experience is key to retaining subscribers. Netflix needed to recommend relevant content to each user from its vast library, ensuring that users remained engaged and satisfied.

Solution: Netflix developed an advanced AI-driven recommendation engine that analyzes individual viewing habits, ratings, and preferences. This personalized approach keeps users engaged, as they are more likely to find content that interests them, enhancing their overall viewing experience.

• Increased viewer engagement and longer watch times.
• Higher subscription retention rates due to personalized content.
• Personalized recommendations significantly enhance user experience.
• AI-driven content curation is essential for success in digital entertainment.

7. Airbus: Elevating Aircraft Maintenance with AI

Task/Conflict: Aircraft maintenance is crucial for ensuring flight safety and operational efficiency. Airbus faced the challenge of predicting maintenance needs to prevent equipment failures and reduce downtime, which is critical in the aviation industry.

Solution: Airbus implemented AI algorithms for predictive maintenance, analyzing data from aircraft sensors to identify potential issues before they lead to failures. This system assesses the condition of various components, predicting when maintenance is needed. The solution not only enhances safety but also optimizes maintenance schedules, reducing unnecessary inspections and downtime.

• Decreased maintenance costs and reduced aircraft downtime.
• Improved safety with proactive maintenance measures.
• AI can predict and prevent potential equipment failures.
• Predictive maintenance is essential for operational efficiency and safety in aviation.

8. American Express: Securing Transactions with AI

Task/Conflict: Credit card fraud is a significant issue in the financial sector, leading to substantial losses and undermining customer trust. American Express needed an efficient way to detect and prevent fraudulent transactions in real-time.

Solution: American Express utilizes machine learning models to analyze transaction data. These models identify unusual patterns and behaviors indicative of fraud. By constant learning from refined data, the system becomes increasingly accurate in detecting fraudulent activities, providing real-time alerts and preventing unauthorized transactions.

• Minimized financial losses due to reduced fraudulent activities.
• Enhanced customer trust and security in financial transactions.
• Machine learning is highly effective in fraud detection.
• Real-time data analysis is crucial for preventing financial fraud.

Related: Is AI a Good Career Option for Women?

9. Stitch Fix: Tailoring the Future of Fashion Retail

Task/Conflict: In the competitive fashion retail industry, providing a personalized shopping experience is key to customer satisfaction and business growth. Stitch Fix aimed to offer customized clothing selections to each customer, based on their unique preferences and style.

Solution: Stitch Fix uses AI and algorithms analyze customer feedback, style preferences, and purchase history to recommend clothing items. This personalized approach is complemented by human stylists, ensuring that each customer receives a tailored selection that aligns with their individual style.

• Increased customer satisfaction through personalized styling services.
• Business growth driven by a unique, AI-enhanced shopping experience.
• AI combined with human judgment can create highly effective personalization.
• Tailoring customer experiences using AI leads to increased loyalty and business success.

10. Baidu: Breaking Language Barriers with Voice Recognition

Task/Conflict: Voice recognition technology faces the challenge of accurately understanding and processing speech in various languages and accents. Baidu aimed to enhance its voice recognition capabilities to provide more accurate and user-friendly interactions in multiple languages.

Solution: Baidu employs deep learning algorithms for voice and speech recognition, training its system on a diverse range of languages and dialects. This approach allows for more accurate recognition of speech patterns, enabling the technology to understand and respond to voice commands more effectively. The system continuously improves as it processes more voice data, making technology more accessible to users worldwide.

• Enhanced user interaction with technology in multiple languages.
• Reduced language barriers in voice-activated services and devices.
• AI can effectively bridge language gaps in technology.
• Continuous learning from diverse data sets is key to improving voice recognition.

11. JP Morgan: Revolutionizing Legal Document Analysis with AI

Task/Conflict: Analyzing legal documents, such as contracts, is a time-consuming and error-prone process. JP Morgan sought to streamline this process, reducing the time and effort required while increasing accuracy.

Solution: JP Morgan implemented an AI-powered tool, COIN (Contract Intelligence), to analyze legal documents quickly and accurately. COIN uses NLP to interpret and extract relevant information from contracts, significantly reducing the time required for document review.

• Dramatic reduction in time required for legal document analysis.
• Increased accuracy and reduced human error in contract interpretation.
• AI can efficiently handle large volumes of data, offering speed and accuracy.
• Automation in legal processes can significantly enhance operational efficiency.

12. Microsoft: AI for Accessibility

Task/Conflict: People with disabilities often face challenges in accessing technology. Microsoft aimed to create AI-driven tools to enhance accessibility, especially for individuals with visual, hearing, or cognitive impairments.

Solution: Microsoft developed a range of AI-powered tools including applications for voice recognition, visual assistance, and cognitive support, making technology more accessible and user-friendly. For instance, Seeing AI, an app developed by Microsoft, helps visually impaired users to understand their surroundings by describing people, texts, and objects.

• Improved accessibility and independence for people with disabilities.
• Creation of more inclusive technology solutions.
• AI can significantly contribute to making technology accessible for all.
• Developing inclusive technology is essential for societal progress.

Related: How to get an Internship in AI?

13. Alibaba’s City Brain: Revolutionizing Urban Traffic Management

Task/Conflict: Urban traffic congestion is a major challenge in many cities, leading to inefficiencies and environmental concerns. Alibaba’s City Brain project aimed to address this issue by using AI to optimize traffic flow and improve public transportation in urban areas.

Solution: City Brain uses AI to analyze real-time data from traffic cameras, sensors, and GPS systems. It processes this information to predict traffic patterns and optimize traffic light timing, reducing congestion. The system also provides data-driven insights for urban planning and emergency response coordination, enhancing overall city management.

• Significant reduction in traffic congestion and improved urban transportation.
• Enhanced efficiency in city management and emergency response.
• AI can effectively manage complex urban systems.
• Data-driven solutions are key to improving urban living conditions.

14. Deep 6 AI: Accelerating Clinical Trials with Artificial Intelligence

Task/Conflict: Recruiting suitable patients for clinical trials is often a slow and cumbersome process, hindering medical research. Deep 6 AI sought to accelerate this process by quickly identifying eligible participants from a vast pool of patient data.

Solution: Deep 6 AI employs AI to sift through extensive medical records, identifying potential trial participants based on specific criteria. The system analyzes structured and unstructured data, including doctor’s notes and diagnostic reports, to find matches for clinical trials. This approach significantly speeds up the recruitment process, enabling faster trial completions and advancements in medical research.

• Quicker recruitment for clinical trials, leading to faster research progress.
• Enhanced efficiency in medical research and development.
• AI can streamline the patient selection process for clinical trials.
• Efficient recruitment is crucial for the advancement of medical research.

15. NVIDIA: Revolutionizing Gaming Graphics with AI

Task/Conflict: Enhancing the realism and performance of gaming graphics is a continuous challenge in the gaming industry. NVIDIA aimed to revolutionize gaming visuals by leveraging AI to create more realistic and immersive gaming experiences.

Solution: NVIDIA’s AI-driven graphic processing technologies, such as ray tracing and deep learning super sampling (DLSS), provide highly realistic and detailed graphics. These technologies use AI to render images more efficiently, improving game performance without compromising on visual quality. This innovation sets new standards in gaming graphics, making games more lifelike and engaging.

• Elevated gaming experiences with state-of-the-art graphics.
• Set new industry standards for graphic realism and performance.
• AI can significantly enhance creative industries, like gaming.
• Balancing performance and visual quality is key to gaming innovation.

16. Palantir: Mastering Data Integration and Analysis with AI

Task/Conflict: Integrating and analyzing large-scale, diverse datasets is a complex task, essential for informed decision-making in various sectors. Palantir Technologies faced the challenge of making sense of vast amounts of data to provide actionable insights for businesses and governments.

Solution: Palantir developed AI-powered platforms that integrate data from multiple sources, providing a comprehensive view of complex systems. These platforms use machine learning to analyze data, uncover patterns, and predict outcomes, assisting in strategic decision-making. This solution enables users to make informed decisions in real-time, based on a holistic understanding of their data.

• Enhanced decision-making capabilities in complex environments.
• Greater insights and efficiency in data analysis across sectors.
• Effective data integration is crucial for comprehensive analysis.
• AI-driven insights are essential for strategic decision-making.

Related: Surprising AI Facts & Statistics

17. Blue River Technology: Sowing the Seeds of AI in Agriculture

Task/Conflict: The agriculture industry faces challenges in increasing efficiency and sustainability while minimizing environmental impact. Blue River Technology aimed to enhance agricultural practices by using AI to make farming more precise and efficient.

Solution: Blue River Technology developed AI-driven agricultural robots that perform tasks like precise planting and weed control. These robots use ML to identify plants and make real-time decisions, such as applying herbicides only to weeds. This targeted approach reduces chemical usage and promotes sustainable farming practices, leading to better crop yields and environmental conservation.

• Significant reduction in chemical usage in farming.
• Increased crop yields through precision agriculture.
• AI can contribute significantly to sustainable agricultural practices.
• Precision farming is key to balancing productivity and environmental conservation.

18. Salesforce: Enhancing Customer Relationship Management with AI

Task/Conflict: In the realm of customer relationship management (CRM), personalizing interactions and gaining insights into customer behavior are crucial for business success. Salesforce aimed to enhance CRM capabilities by integrating AI to provide personalized customer experiences and actionable insights.

Solution: Salesforce incorporates AI-powered tools into its CRM platform, enabling businesses to personalize customer interactions, automate responses, and predict customer needs. These tools analyze customer data, providing insights that help businesses tailor their strategies and communications. The AI integration not only improves customer engagement but also streamlines sales and marketing efforts.

• Improved customer engagement and satisfaction.
• Increased business growth through tailored marketing and sales strategies.
• AI-driven personalization is key to successful customer relationship management.
• Leveraging AI for data insights can significantly impact business growth.

19. OpenAI: Transforming Natural Language Processing

Task/Conflict: OpenAI aimed to advance NLP by developing models capable of generating coherent and contextually relevant text, opening new possibilities in AI-human interaction.

Solution: OpenAI developed the Generative Pre-trained Transformer (GPT) models, which use deep learning to generate text that closely mimics human language. These models are trained on vast datasets, enabling them to understand context and generate responses in a conversational and coherent manner.

• Pioneered advancements in natural language understanding and generation.
• Expanded the possibilities for AI applications in communication.
• AI’s ability to mimic human language has vast potential applications.
• Advancements in NLP are crucial for improving AI-human interactions.

20. Siemens: Pioneering Industrial Automation with AI

Task/Conflict: Industrial automation seeks to improve productivity and efficiency in manufacturing processes. Siemens faced the challenge of optimizing these processes using AI to reduce downtime and enhance output quality.

Solution: Siemens employs AI-driven solutions for predictive maintenance and process optimization to reduce downtime in industrial settings. Additionally, AI optimizes manufacturing processes, ensuring quality and efficiency.

• Increased productivity and reduced downtime in industrial operations.
• Enhanced quality and efficiency in manufacturing processes.
• AI is a key driver in the advancement of industrial automation.
• Predictive analytics are crucial for maintaining efficiency in manufacturing.

Related: Top Books for Learning AI

Closing Thoughts

These 20 case studies illustrate the transformative power of AI across various industries. By addressing specific challenges and leveraging AI solutions, companies have achieved remarkable outcomes, from enhancing customer experiences to solving complex scientific problems. The key learnings from these cases underscore AI’s potential to revolutionize industries, improve efficiencies, and open up new possibilities for innovation and growth.

• 6 Ways to Pay for an Online Course [2024]
• How to Choose an Online Course? [An Ultimate Checklist] [2024]

Team DigitalDefynd

We help you find the best courses, certifications, and tutorials online. Hundreds of experts come together to handpick these recommendations based on decades of collective experience. So far we have served 4 Million+ satisfied learners and counting.

How Can a CFO Use ChatGPT? [2024]

What Teaching Jobs Are Safe from AI and Automation? [2024]

How is AI Empowering the Electric Car Industry? [2024]

Top 60 AI Terms Defined [2024]

Ten Examples of AI Being Used in the Navy [2024]

15 Ways AI Is Helping Humanity [2024]

100+ AI Use Cases & Applications: In-Depth Guide for 2024

AI is changing every industry and business function, which results in increased interest in AI, its subdomains, and related fields such as machine learning and data science as seen below. With the launch of ChatGPT , interest in generative AI , a subfield of AI, exploded.

According to a recent McKinsey survey, 55% of organizations are using AI in at least one business function. 1 To integrate AI into your own business, you need to identify how AI can serve your business, possible use cases of AI in your business.

This article gathers the most common AI use cases covering marketing, sales, customer services, security, data, technology, and other processes.

Generative AI Use Cases

Generative AI involves AI models generating output in requests where there is not a single right answer (e.g. creative writing). Since the launch of ChatGPT , it has been exploding in popularity. Its use cases include content creation for marketing, software code generation, user interface design and many others.

For more: Generative AI use cases .

Business Functions

> ai use cases for analytics, general solutions.

• Analytics Platform : Empower your employees with unified data and tools to run advanced analyses. Quickly identify problems and provide meaningful insights.
• Analytics Services : Satisfy your custom analytics needs with these e2e solution providers. Vendors are there to help you with your business objectives by providing turnkey solutions.
• Automated Machine Learning (autoML) : Machines helping data scientists optimize machine learning models. With the rise of data and analytics capabilities, automation is needed in data science. AutoML automates time consuming machine learning tasks, enabling companies to deploy models and automate processes faster.

Specialized solutions

• Conversational Analytics : Use conversational interfaces to analyze your business data. Natural Language Processing is there to help you with voice data and more enabling automated analysis of reviews and suggestions.
• E-Commerce Analytics : Specialized analytics systems designed to deal with the explosion of e-commerce data. Optimize your funnel and customer traffic to maximize your profits.
• Geo-Analytics Platform : Enables analysis of granular satellite imagery for predictions. Leverage spatial data for your business goals. Capture the changes in any landscape on the fly.
• Image Recognition and Visual Analytics : Analyze visual data with advanced image and video recognition systems. Meaningful insights can be derived from the data piles of images and videos.
• Real-Time Analytics : Real-Time Analytics for your time-sensitive decisions. Act timely and keep your KPI’s intact. Use machine learning to explore unstructured data without any disruptions.

> AI use cases for Customer Service

• Call Analytics : Advanced analytics on call data to uncover insights to improve customer satisfaction and increase efficiency. Find patterns and optimize your results. Analyze customer reviews through voice data and pinpoint, where there is room for improvement. Sestek indicates that ING Bank observed a 15% increase in sales quality score and a 3% decrease in overall silence rates after they integrated AI into their contact systems .
• Call Classification : Leverage natural language processing (NLP) to understand what the customer wants to achieve so your agents can focus on higher value-added activities. Before channeling the call, identify the nature of your customers’ needs and let the right department handle the problem. Increase efficiency with higher satisfaction rates.
• Call Intent Discovery : Leverage Natural Language Processing and machine learning to estimate and manage customer’s intent (e.g., churn) to improve customer satisfaction and business metrics. Sentiment analysis through the customer’s voice level and pitch. Detect the micro-emotions that drive the decision-making process. Explore how chatbots detect customer intent in our in-depth article on intent recognition .
• Chatbot for Customer Service (Self – Service Solution) : Chatbots can understand more complicated queries as AI algorithms improve. Build your own 24/7 functioning, intelligent, self-improving chatbots to handle most queries and transfer customers to live agents when needed. Reduce customer service costs and increase customer satisfaction. Reduce the traffic on your existing customer representatives and make them focus on the more specific needs of your customers. Read for more insights on chatbots in customer service or discover chatbot platforms .
• Chatbot Analytics : Analyze how customers are interacting with your chatbot. See the overall performance of your chatbot. Pinpoint its shortcomings and improve your chatbot. Detect the overall satisfaction rate of your customer with the chatbot.
• Chatbot testing : Semi-automated and automated testing frameworks facilitate bot testing. See the performance of your chatbot before deploying. Save your business from catastrophic chatbot failures. Detect the shortcomings of your conversational flow.
• Customer Contact Analytics : Advanced analytics on all customer contact data to uncover insights to improve customer satisfaction and increase efficiency. Utilize Natural Language Processing for higher customer satisfaction rates.
• Customer Service Response Suggestions : Bots will listen in on agents’ calls suggesting best practice answers to improve customer satisfaction and standardize customer experience. Increase upsells and cross-sells by giving the right suggestion. Responses will be standardized, and the best possible approach will serve the benefit of the customer.
• Social Listening & Ticketing : Leverage Natural Language Processing and machine vision to identify customers to contact and respond to them automatically or assign them to relevant agents, increasing customer satisfaction. Use the data available in social networks to uncover whom to sell and what to sell.
• Intelligent Call Routing : Route calls to the most capable agents available. Intelligent routing systems incorporate data from all customer interactions to optimize the customer satisfaction. Based on the customer profile and your agent’s performance, you can deliver the right service with the right agent and achieve superior net promoter scores. Feel free to read case studies about matching customer to right agent in our emotional AI examples article .
• Survey & Review Analytics : Leverage Natural Language Processing to analyze text fields in surveys and reviews to uncover insights to improve customer satisfaction and increase efficiency. Automate the process by mapping the right keywords with the right scores. Make it possible to lower the time for generating reports. Protobrand states that they used to do review analytics manually through the hand-coding of the data, but now it automates much of the analytical work with Gavagai. This helps the company to collect larger quantitative volumes of qualitative data and still complete the analytical work in a timely and efficient manner. You can read more about survey analytics from  our related article .
• Voice Authentication : Authenticate customers without passwords leveraging biometry to improve customer satisfaction and reduce issues related to forgotten passwords. Their unique voice id will be their most secure key for accessing confidential information. Instead of the last four digits of SSN, customers will gain access by using their voice.

> AI use cases for Data

• Data Cleaning & Validation Platform : Avoid garbage in, garbage out by ensuring the quality of your data with appropriate data cleaning processes and tools. Automate the validation process by using external data sources. Regular maintenance cleaning can be scheduled, and the quality of the data can be increased.
• Data Integration : Combine your data from different sources into meaningful and valuable information. Data traffic depends on multiple platforms. Therefore, managing this huge traffic and structuring the data into a meaningful format will be important. Keep your data lake available for further analysis. 
• Data Management & Monitoring : Keep your data high quality for advanced analytics. Adjust the quality by filtering the incoming data. Save time by automating manual and repetitive tasks.
• Data Preparation Platform : Prepare your data from raw formats with data quality problems to a clean, ready-to-analyze format. Use extract, transform, and load (ETL) platforms to fine-tune your data before placing it into a data warehouse.
• Data Transformation : Transform your data to prepare it for advanced analytics. If it is unstructured, adjust it for the required format.
• Data Visualization : Visualize your data for better analytics and decision-making. Let the dashboards speak. Convey your message more easily and more esthetically.
• Data Labeling : Unless you use unsupervised learning systems, you need high quality labeled data. Label your data to train your supervised learning systems. Human-in-the-loop systems auto label your data and crowdsource labeling data points that cannot be auto-labeled with confidence.
• Synthetic Data :  Computers can artificially create synthetic data to perform certain operations. The synthetic data is usually used to test new products and tools, validate models, and satisfy AI needs. Companies can simulate not yet encountered conditions and take precautions accordingly with the help of synthetic data. They also overcome the privacy limitations as it doesn’t expose any real data. Thus, synthetic data is a smart AI solution for companies to simulate future events and consider future possibilities. You can have more information on synthetic data from  our related article .

> AI use cases for Finance

Finance business function led by the CEO completes numerous repetitive tasks involving quantitative skills which makes them a good fit for AI transformation:

• Billing / invoicing reminders : Leverage accessible billing services that remind your customers to pay.
• Invoicing : Invoicing is a highly repetitive process that many companies perform manually. This causes human errors in invoicing and high costs in terms of time, especially when a high volume of documents needs to be processed. Thus, companies can handle these repetitive tasks with AI, automate invoicing procedures, and save significant time while reducing invoicing errors.

> AI use cases for HR

• Employee Monitoring : Monitor your employees for better productivity measurement. Provide objective metrics to see how well they function. Forecast their overall performance with the availability of massive amounts of data.
• Hiring :  Hiring is a prediction game: Which candidate, starting at a specific position, will contribute more to the company? Machine and recruiting chatbots ‘ better data processing capabilities augment HR employees in various parts of hiring such as finding qualified candidates, interviewing them with bots to understand their fit or evaluating their assessment results to decide if they should receive an offer. 
• HR Analytics : HR analytics services are like the voice of employee analysis. Look at your workforce analytics and make better HR decisions. Gain actionable insights and impactful suggestions for higher employee satisfaction.
• HR Retention Management : Predict which employees are likely to churn and improve their job satisfaction to retain them. Detect the underlying reasons for their motive for seeking new opportunities. By keeping them at your organization, lower your human capital loss.
• Performance Management : Manage your employees’ performance effectively and fairly without hurting their motivation. Follow their KPI’s on your dashboard and provide real-time feedback. This would increase employee satisfaction and lower your organization’s employee turnover. Actualize your employee’s maximum professional potential with the right tools.

You can also read our article on HR technology trends .

> AI use cases for Marketing

A 2021 survey conducted among global marketers revealed that 41% of respondents saw an increase in revenue growth and improved performance due to the use of AI in their marketing campaigns.

Marketing can be summarized as reaching the customer with the right offer, the right message, at the right time, through the right channel, while continually learning. To achieve success, companies can leverage AI-powered tools to get familiar with their customers better, create more compelling content, and perform personalized marketing campaigns. AI can provide accurate insights and suggest smart marketing solutions that would directly reflect on profits with customer data. You can find the top three AI use cases in marketing:

• Marketing analytics :  AI systems learn from, analyze, and measure marketing efforts. These solutions track media activity and provide insights into PR efforts to highlight what is driving engagement, traffic, and revenue. As a result, companies can provide better and more accurate marketing services to their customers. Besides PR efforts, AI-powered marketing analytics can lead companies to identify their customer groups more accurately. By discovering their loyal customers, companies can develop accurate marketing strategies and also retarget customers who have expressed interest in products or services before. Feel free to read more about marketing analytics with AI from  this article .
• Personalized Marketing:  The more companies understand their customers, the better they serve them. AI can assist companies in this task and support them in giving personalized experiences for customers. As an example, suppose you visited an online store and looked at a product but didn’t buy it. Afterward, you see that exact product in digital ads. More than that, companies can send personalized emails or special offers and recommend new products that go along with customers’ tastes.
• Context-Aware Marketing : You can leverage machine vision and natural language processing (NLP) to understand the context where your ads will be served. With context-aware advertising, you can protect your brand and increase marketing efficiency by ensuring your message fits its context, making static images on the web come alive with your messages. 

To learn more about AI use cases in marketing, you can check out  our complete guide  on the topic.

> AI use cases for Operations

• Cognitive / Intelligent Automation : Combine robotic process automation (RPA) with AI to automate complex processes with unstructured information. Digitize your processes in weeks without replacing legacy systems , which can take years. Bots can operate on legacy systems learning from your personnel’s instructions and actions. Increase your efficiency and profitability ratios. Increase speed and precision, and many more. Feel free to check intelligent automation use cases for more.
• Robotic Process Automation (RPA) Implementation : Implementing RPA solutions requires effort. Suitable processes need to be identified. If a rules-based robot will be used, the robot needs to be programmed. Employees’ questions need to be answered. That is why most companies get some level of external help. Generally, outsourcing companies, consultants, and IT integrators are happy to provide temporary labor to undertake this effort.
• Process Mining : Leverage AI algorithms to mine your processes and understand your actual processes in detail. Process mining tools can provide fastest time to insights about your as-is processes as demonstrated in case studies . Check out process mining use cases & benefits for more.
• Predictive Maintenance : Predictively maintain your robots and other machinery to minimize disruptions to operations. Implement big data analytics to estimate the factors that are likely to impact your future cash flow. Optimize PP&E spending by gaining insight regarding the possible factors.
• Inventory & Supply Chain Optimization : Leverage machine learning to take your inventory& supply chain optimization to the next level. See the possible scenarios in different customer demands. Reduce your stock, keeping spending, and maximize your inventory turnover ratios. Increase your impact factor in the value chain.
• Building Management : Sensors and advanced analytics improve building management. Integrate IoT systems in your building for lower energy consumption and many more. Increase the available data by implementing the right data collection tools for effective building management.
• Digital Assistant : Digital assistants are mature enough to replace real assistants in email communication. Include them in your emails to schedule meetings. They have already scheduled hundreds of thousands of meetings.

> AI use cases for Sales

• Sales Forecasting :  AI allows automatic and accurate sales forecasts based on all customer contacts and previous sales outcomes. Automatically forecast sales accurately based on all customer contacts and previous sales outcomes. Give your sales personnel more sales time while increasing forecast accuracy. Hewlett Packard Enterprise indicates that it has experienced a 5x increase in forecast simplicity, speed, and accuracy with Clari’s sales forecasting tools.
• Lead generation :  Use a comprehensive data profile of your visitors to identify which companies your sales reps need to connect. Generate leads for your sales reps leveraging databases and social networks
• Sales Data Input Automation: Data from various sources will be effortlessly and intelligently copied into your CRM. Automatically sync calendar, address book, emails, phone calls, and messages of your salesforce to your CRM system. Enjoy better sales visibility and analytics while giving your sales personnel more sales time.
• Predictive sales/lead scoring: Use AI to enable predictive sales. Score leads to prioritize sales rep actions based on lead scores and contact factors. Sales forecasting is automated with increased accuracy thanks to systems’ granular access to lead scores and sales rep performance. For scoring leads, these systems leverage anonymized transaction data from their customers, sales data of this specific customer. For assessing contact factors, these systems leverage anonymized data and analyze all customer contacts such as email and calls.
• Sales Rep Response Suggestions: AI will suggest responses during live conversations or written messages with leads. Bots will listen in on agents’ calls suggesting best practice answers to improve sales effectiveness
• Sales Rep Next Action Suggestions : Your sales reps’ actions and leads will be analyzed to suggest the next best action. This situation wise solution will help your representatives to find the right way to deal with the issue. Historical data and profile of the agent will help you to achieve higher results. All are leading to more customer satisfaction.
• Sales Content Personalization and Analytics: Preferences and browsing behavior of high priority leads are analyzed to match them with the right content, aimed to answer their most important questions. Personalize your sales content and analyze its effectiveness allowing continuous improvement.
• Retail Sales Bot : Use bots on your retail floor to answer customer’s questions and promote products. Engage with the right customer by analyzing the profile. Computer vision will help you to provide the right action depending on the characteristics and mimics of the customer.
• Meeting Setup Automation (Digital Assistant): Leave a digital assistant to set up meetings freeing your sales reps time. Decide on the targets to prioritize and keep your KPI’s high.
• Prescriptive Sales : Most sales processes exist in the mind of your sales reps. Sales reps interact with customers based on their different habits and observations. Prescriptive sales systems prescribe the content, interaction channel, frequency, price based on data on similar customers .
• Sales Chatbot : Chatbots are ideal to answer first customer questions. If the chatbot decides that it can not adequately serve the customer, it can pass those customers to human agents. Let 24/7 functioning, intelligent, self-improving bots handle making initial contacts to leads. High value, responsive leads will be called by live agents, increasing sales effectiveness.

Sales analytics

As Gartner discusses , sales analytic systems provide functionality that supports discovery, diagnostic, and predictive exercises that enable the manipulation of parameters, measures, dimensions, or figures as part of an analytic or planning exercise. AI algorithms can automate the data collection process and present solutions to improve sales performance. To have more detailed information, you can read  our article about sales analytics .

• Customer Sales Contact Analytics :  Analyze all customer contacts, including phone calls or emails, to understand what behaviors and actions drive sales. Advanced analytics on all sales call data to uncover insights to increase sales effectiveness
• Sales Call Analytics : Advanced analytics on call data to uncover insights to increase sales effectiveness. See how well your conversation flow performs. Integrating data on calls will help you to identify the performance of each component in your sales funnels.
• Sales attribution :  Leverage big data to attribute sales to marketing and sales efforts accurately. See which step of your sales funnel performs better. Pinpoint the low performing part by the insights provided by analysis.
• Sales Compensation :  Determine the right compensation levels for your sales personnel. Decide on the right incentive mechanism for the sales representatives. By using the sales data, provide objective measures, and continuously increase your sales representatives’ performance.

For more on AI in sales .

> AI use cases for Tech

• No code AI & app development : AI and App development platforms for your custom projects. Your in-house development team can create original solutions for your specific business needs.
• Analytics & Predictive Intelligence for Security : Analyze data feeds about the broad cyber activity as well as behavioral data inside an organization’s network to come up with actionable insights to help analysts predict and thwart impending attacks. Integrate external data sources the watch out for global cyber threats and act timely. Keep your tech infrastructure intact or minimize losses. 
• Knowledge Management : Enterprise knowledge management enables effective and effortless storage and retrieval of enterprise data, ensuring organizational memory. Increased collaboration by ensuring the right people are working with the right data. Seamless organizational integration through knowledge management platforms.
• Natural Language Processing Library/ SDK/ API : Leverage Natural Language Processing libraries/SDKs/APIs to quickly and cost-effectively build your custom NLP powered systems or to add NLP capabilities to your existing systems. An in-house team will gain experience and knowledge regarding the tools. Increased development and deployment capabilities for your enterprise.
• Image Recognition Library/ SDK/ API :  Leverage image recognition libraries/SDKs/APIs to quickly and cost-effectively build your custom image processing systems or to add image processing capabilities to your existing systems.
• Secure Communications : Protect employee communications like emails or phone conversations with advanced multilayered cryptography & ephemerality. Keep your industry secrets safe from corporate espionage.
• Deception Security : Deploy decoy-assets in a network as bait for attackers to identify, track, and disrupt security threats such as advanced automated malware attacks before they inflict damage. Keep your data and traffic safe by keeping them engaged in decoys. Enhance your cybersecurity capabilities against various forms of cyber attacks
• Autonomous Cybersecurity Systems : Utilize learning systems to efficiently and instantaneously respond to security threats, often augmenting the work of security analysts. Lower your risk of human errors by providing greater autonomy for your cybersecurity. AI-backed systems can check compliance with standards.
• Smart Security Systems : AI-powered autonomous security systems. Functioning 24/7 for achieving maximum protection. Computer vision for detecting even the tiniest anomalies in your environment. Automate emergency response procedures by instant notification capabilities.
• Machine Learning Library/ SDK/ API : Leverage machine learning libraries/SDKs/APIs to quickly and cost-effectively build your custom learning systems or to add learning capabilities to your existing systems.
• AI Developer : Develop your custom AI solutions with companies experienced in AI development. Create turnkey projects and deploy them to the specific business function. Best for companies with limited in-house capabilities for artificial intelligence.
• Deep Learning Library/ SDK/ API : Leverage deep learning libraries/SDKs/APIs to quickly and cost-effectively build your custom learning systems or to add learning capabilities to your existing systems.
• Developer Assistance : Assist your developers using AI to help them intelligently access the coding knowledge on the web and learn from suggested code samples. See the best practices for specific development tasks and formulate your custom solution. Real-time feedback provided by the huge history of developer mistakes and best practices.
• AI Consultancy : Provides consultancy services to support your in-house AI development, including machine learning and data science projects. See which units can benefit most from AI deployment. Optimize your artificial intelligence spending for the best results from the insight provided by a consultant.

> AI use cases for Automotive & Autonomous Things

Autonomous things including cars and drones are impacting every business function from operations to logistics.

• Driving Assistant : Required components and intelligent solutions to improve rider’s experience in the car. Implement AI-Powered vehicle perception solutions for the ultimate driving experience.
• Vehicle Cybersecurity : Secure connected and autonomous cars and other vehicles with intelligent cybersecurity solutions. Guarantee your safety by hack-proof mechanisms. Protect your intelligent systems from attacks.
• Vision Systems : Vision systems for self-driving cars. Integrate vision sensing and processing in your vehicle. Achieve your goals with the help of computer vision.
• Self-Driving Cars : From mining to manufacturing, self-driving cars/vehicles are increasing the efficiency and effectiveness of operations. Integrate them into your business for greater efficiency. Leverage the power of artificial intelligence for complex tasks.

> AI use cases for Education

• Course creation

For more: Generative AI applications in education

> AI use cases for Fashion

• Creative Design
• Virtual try-on
• Trend analysis

For more: Generative AI applications in fashion

> AI use cases for FinTech 

• Fraud Detection : Leverage machine learning to detect fraudulent and abnormal financial behavior, and/or use AI to improve general regulatory compliance matters and workflows. Lower your operational costs by limiting your exposure to fraudulent documents.
• Insurance & InsurTech : Leverage machine learning to process underwriting submissions efficiently and profitably, quote optimal prices , manage claims effectively, and improve customer satisfaction while reducing costs. Detect your customer’s risk profile and provide the right plan.
• Financial Analytics Platform : Leverage machine learning, Natural Language Processing, and other AI techniques for financial analysis, algorithmic trading, and other investment strategies or tools.
• Travel & expense management : Use deep learning to improve data extraction from receipts of all types including hotel, gas station, taxi, grocery receipts. Use anomaly detection and other approaches to identify fraud, non-compliant spending. Reduce approval workflows and processing costs per unit.
• Credit Lending & Scoring : Use AI for robust credit lending applications. Use predictive models to uncover potentially non-performing loans and act. See the potential credit scores of your customers before they apply for a loan and provide custom-tailored plans.
• Loan recovery: Increase loan recovery ratios with empathetic and automated messages.
• Robo-Advisory : Use AI finance chatbot and mobile app assistant applications to monitor personal finances. Set your target savings or spending rates for your own goals. Your finance assistant will handle the rest and provide you with insights to reach financial targets.
• Regulatory Compliance : Use Natural Language Processing to quickly scan legal and regulatory text for compliance issues, and do so at scale. Handle thousands of paperwork without any human interaction.
• Data Gathering : Use AI to efficiently gather external data such as sentiment and other market-related data. Wrangle data for your financial models and trading approaches.
• Debt Collection : Leverage AI to ensure a compliant and efficient debt collection process. Effectively handle any dispute and see your success right in debt collection.
• Conversational banking : Financial institutions engage with their customers on a variety of communication platforms ( WhatsApp , mobile app , website etc.) via conversational AI tools to increase customer satisfaction and automate many tasks like customer onboarding .

> AI use cases for HealthTech

• Patient Data Analytics : Analyze patient and/or 3rd party data to discover insights and suggest actions. Greater accuracy by assisted diagnostics. Lower the mortality rates and increase patient satisfaction by using all the diagnostic data available to detect the underlying reasons for the symptoms.
• Personalized Medications and Care : Find the best treatment plans according to patient data. Provide custom-tailored solutions for your patients. By using their medical history, genetic profile, you can create a custom medication or care plan.
• Drug Discovery : Find new drugs based on previous data and medical intelligence. Lower your R&D cost and increase the output — all leading to greater efficiency. Integrate FDA data, and you can transform your drug discovery by locating market mismatches and FDA approval or rejection rates.
• Real-Time Prioritization and Triage : Prescriptive analytics on patient data enabling accurate real-time case prioritization and triage. Manage your patient flow by automatization. Integrate your call center and use language processing tools to extract the information, priorate patients that need urgent care, and lower your error rates. Eliminate error-prone decisions by optimizing patient care.
• Early Diagnosis : Analyze chronic conditions leveraging lab data and other medical data to enable early diagnosis. Provide a detailed report on the likelihood of the development of certain diseases with genetic data. Integrate the right care plan for eliminating or reducing the risk factors.
• Assisted or Automated Diagnosis & Prescription :  Suggest the best treatment based on the patient complaint and other data. Put in place control mechanisms that detect and prevent possible diagnosis errors. Find out which active compound is most effective against that specific patient. Get the right statistics for superior care management.
• Pregnancy Management : Monitor mother and fetus health to reduce mothers’ worries and enable early diagnosis. Use machine learning to uncover potential risks and complications quickly. Lower the rates of miscarriage and pregnancy-related diseases.
• Medical Imaging Insights : Advanced medical imaging to analyze and transform images and model possible situations. Use diagnostic platforms equipped with high image processing capabilities to detect possible diseases.
• Healthcare Market Research : Prepare hospital competitive intelligence by tracking market prices. See the available insurance plans, drug prices, and many more public data to optimize your services. Leverage NLP tools to analyze the vast size of unstructured data.
• Healthcare Brand Management and Marketing : Create an optimal marketing strategy for the brand based on market perception and target segment. Tools that offer high granularity will allow you to reach the specific target and increase your sales.
• Gene Analytics and Editing : Understand genes and their components and predict the impact of gene edits.
• Device and Drug Comparative Effectiveness : Analyze drug and medical device effectiveness. Rather than just using simulations, test on other patient’s data to see the effectiveness of the new drug, compare your results with benchmark drugs to make an impact with the drug.
• Healthcare chatbot :  Use a chatbot to schedule patient appointments, give information about certain diseases or regulations, fill in patient information, handle insurance inquiries, and provide mental health assistance. You can also use intelligent automation with chatbot capabilities.

For more, feel free to check our article on the  use cases of AI in the healthcare industry .

> AI use cases for Manufacturing

• Manufacturing Analytics : Also called industrial analytics systems, these systems allow you to analyze your manufacturing process from production to logistics to save time, reduce cost, and increase efficiency. Keep your industry effectiveness at optimal levels.
• Collaborative Robots : Cobots provide a flexible method of automation. Cobots are flexible robots that learn by mimicking human workers’ behavior.
• Robotics : Factory floors are changing with programmable collaborative bots that can work next to employees to take over more repetitive tasks. Automate physical processes such as manufacturing or logistics with the help of advanced robotics. Increased your connected systems by centralizing the whole manufacturing process. Lower your exposures to human errors.

> AI use cases for Retail

• Cashierless Checkout : Self-checkout systems have many names. They are called cashierless, cashier-free, or automated checkout systems. They allow retail companies to serve customers in their physical stores without the need for cashiers. Technologies that allowed users to scan and pay for their products have been used for almost a decade now, and those systems did not require great advances in AI. However, these days we are witnessing systems powered by advanced sensors and AI to identify purchased merchandise and charge customers automatically.

> AI use cases for Telecom

• Network investment optimization : Both wired and wireless operators need to invest in infrastructure like active equipment or higher bandwidth connections to improve Quality of Service (QoS). Machine learning can be used to identify highest ROI investments that will result in less churn and higher cross and up-sell.

Other AI Use Cases

This was a list of areas by business function where out-of-the-box solutions are available. However, AI, like software, has too many applications to list here. You can also take a look at our  AI in business article  to read about AI applications by industry. Also, feel free to check our article on AI services .

It is important to get started fast with high impact applications and generate business value without spending months of effort. For that, we recommend companies to use no code AI solutions to quickly build AI models .

Once companies deploy a few models to production, they need to take a deeper look at their AI/ML development model.

• rely on autoML software to build complex AI models. Though most autoML software is not as easy to use as no code AI solutions, they can be used to build complex models.
• build custom AI solutions in-house
• work with the support of partners to build custom models
• run data science competitions to build custom AI models
• Use pre-trained models built by AI vendors

We examined the pros and cons of this approaches in our article on making the build or buy decisions regarding AI .

You can also check out our list of AI tools and services:

• AI Consultant
• AI/ML Development Services
• Data Science / ML / AI Platform

These articles about AI may also interest you:

• Ultimate Guide to the State of AI technology
• Future of AI according to top AI experts
• Advantages of AI according to top practitioners

What is artificial intelligence (AI)?

Artificial Intelligence (AI) is the branch of computer science that focuses on creating machines capable of performing tasks that typically require human intelligence. This includes activities such as learning, problem-solving, understanding natural language, speech recognition, and visual perception. AI systems can analyze large amounts of data, identify patterns, and make decisions, often with speed and accuracy surpassing human capabilities.

What are the examples of AI in real life?

Artificial Intelligence (AI) is integrated into many aspects of daily life. Some common real-life examples include:

Virtual Assistants: Like Siri, Alexa, and Google Assistant, these AI-powered tools understand and respond to voice commands, performing tasks like setting reminders, answering questions, and controlling smart home devices.

Navigation and Maps: AI is used in services like Google Maps and Waze for route optimization, traffic prediction, and providing real-time directions.

Recommendation Systems: Streaming services like Netflix and Spotify use AI to analyze your viewing or listening history to recommend movies, shows, or music.

Autonomous Vehicles: Self-driving cars use AI to perceive the environment and make decisions for safe navigation.

Social Media: Platforms like Facebook and Instagram use AI for content curation, targeted advertising, and facial recognition in photos.

Security and Surveillance: AI aids in anomaly detection, facial recognition, and monitoring systems for enhanced security.

How does AI impact employment and job creation?

AI impacts employment by automating routine tasks, which can lead to job displacement in some sectors. However, it also creates new job opportunities in AI development, data analysis, and other tech-related fields, emphasizing the need for skill adaptation.

For more, you can check our article on the ethics of AI .

What are some misconceptions about AI?

Common misconceptions include the idea that AI can fully replicate human intelligence, that it’s always unbiased, or that AI-led automation will universally eliminate jobs. In reality, AI has limitations, can inherit biases from data, and often changes rather than replaces job roles.

And if you have a specific business challenge, we can help you find the right vendor to overcome that challenge:

External links

Though most use cases have been categorized based on our experience, we also took a look at Tractica’s AI use cases list before finalizing the list. Other sources:

• 1. “ The state of AI in 2023: Generative AI’s breakout year “. Quantum Black AI by McKinsey . August 1, 2023. Accessed January 1, 2024

Cem has been the principal analyst at AIMultiple since 2017. AIMultiple informs hundreds of thousands of businesses (as per similarWeb) including 60% of Fortune 500 every month. Cem's work has been cited by leading global publications including Business Insider , Forbes, Washington Post , global firms like Deloitte , HPE, NGOs like World Economic Forum and supranational organizations like European Commission . You can see more reputable companies and media that referenced AIMultiple. Throughout his career, Cem served as a tech consultant, tech buyer and tech entrepreneur. He advised businesses on their enterprise software, automation, cloud, AI / ML and other technology related decisions at McKinsey & Company and Altman Solon for more than a decade. He also published a McKinsey report on digitalization. He led technology strategy and procurement of a telco while reporting to the CEO. He has also led commercial growth of deep tech company Hypatos that reached a 7 digit annual recurring revenue and a 9 digit valuation from 0 within 2 years. Cem's work in Hypatos was covered by leading technology publications like TechCrunch and Business Insider . Cem regularly speaks at international technology conferences. He graduated from Bogazici University as a computer engineer and holds an MBA from Columbia Business School.

To stay up-to-date on B2B tech & accelerate your enterprise:

Next to Read

Ai center of excellence (ai coe): what it is & how to build in '24, ai chips: a guide to cost-efficient ai training & inference in 2024, ai in analytics: how ai is shaping analytics in 2024 in 4 ways.

Your email address will not be published. All fields are required.

Good afternoon. I am very curious about your claim that “Elekta has reduced its costs and increased its number of processed invoices from 50,000 to 120,000.” Do you have the source for this claim?

Hello, Aidan. We weren’t able to find the source. So we removed it entirely. Thanks for pointing it out!

We can say that AI is the future of our world. While AI is penetrating in more and more human works, thus creating a demand of AI Industry, AI in healthcare is one of the most surging category in global AI Market. According to Meridian Market Consultants, The global AI in Healthcare Market in 2020 is estimated for more than US$ 5.0 Bn and expected to reach a value of US$ 107.5 Bn by 2028 with a significant CAGR of 47.3%. SOI:

47.3% CAGR? You are so sure about the future. Why don’t you guys just sell the time machine rather than the report?

Related research

Vertical AI / Horizontal AI & Other Specialized AI Models in 2024

AI in Marketing: Comprehensive Guide in 2024

Best AI Case Study Examples in 2024 (And a How-To Guide!)

Who has the best case studies for ai solutions.

B2B buyers’ heads are spinning with the opportunities that AI makes possible.

But in a noisy, technical space where hundreds of new AI solutions and use cases are popping up overnight, many buyers don’t know how to navigate these opportunities—or who they can trust.

Your customers are as skeptical as they are excited, thinking…

• “I’m confused by the complexity of your technology.”
• “I’m unsure whether there’s clear ROI.”
• “I’m concerned about my data security.”
• “How will I integrate AI into our systems?”
• “I’m worried about employee pushback.”
• “I’m nervous about its use and governance.”

Done well, case studies about your AI solution can answer all of these questions in a way no other asset can:

With real-world storytelling, third party trust, and practical demonstrations that you can do what you promise.

To help you level up your customer stories, we’ve scoured the web for examples of the best AI case studies from companies spanning billion-dollar-juggernauts and scrappy startups.

Then, we profiled exactly what they’re doing well so you can level up your own stories!

OPPORTUNITY ALERT: Of all of the businesses we reviewed in researching this piece, just 50% were publishing customer success stories on their websites. Want an instant competitive advantage in AI? Scale your own case study production right now!

1. Location is everything: make stories findable

Key decision-makers in B2B businesses actively seek out word-of-mouth content about potential AI partners (like you!). So the easier they can find case studies on your website, the better.

Of the AI businesses we analyzed doing case studies, most make it easy to locate their case study overview page (where prospects see your complete portfolio of case studies at a glance.)

A common journey is via ‘Resources’ in the main navigation bar, followed by a link to ‘Customer Stories’, ‘Client Stories’, ‘Case Studies’, or similar.

For example, Otter.ai has their customer stories slightly buried in their “Blog” section , with an easy-to-miss category link. We don’t love this, because there’s no clear reason someone should expect to find this type of content in the blog vs. a “Customers” section or otherwise:

These also appear in their “Resources” section, but without any sort of jump link or clear indication you might find them there:

But you can do better!

In a space so skeptical and noisy, we advise you follow the likes of Presight AI and Google DeepMind and give buyers access to your customer success stories with a single click from the main navigation:

While Presight AI favors simplicity with a link to ‘Client Stories’, Google DeepMind opens the door via ‘Impact’.

If, like Google DeepMind, your impact as an AI business extends beyond commercial customers to broader sectors and communities, using a term like ‘Impact’ works well, but ‘clear’ is better than clever here, and a simpler term (‘Customers’) may be stronger.

You’ve put in the hard work sourcing concrete proof for potential buyers; don’t put hurdles in the way of finding it.

AI case study overview pages

The ‘overview’ for your customer story page is where customers are going to either continue their journey with intention—or stumble around in the dark.

A great overview page provides a clear sense of hierarchy (what’s important?), organization (what’s here, and what’s for me?) and expectation (what’s on the other side of the click?).

Take Jasper.ai for example:

Their overview page starts strong with a compelling bit of social proof (100,000+ businesses? Holy toledo!). Having a featured story is great (more on that later), though the headline for the one in the image sort of buries the lede (800% surge in traffic!? Holy toledo!)

After that, the page offers no clear way to drill down with intention: A lead is left to scroll through the logos presented to see if there are any companies they know of, or choose a story at random—most likely the featured story or the one in the upper left of the grid.

That’s not as ideal: you’d much rather have a customer quickly find the stories most relevant to THEM.

Boston Dynamics is one AI business worth emulating on that front.

A no-nonsense intro tells prospects they’re on the right page: “Discover success stories from real customers putting our robotics systems to work.”

If you choose to run a featured case study on your overview page, choose a high-impact one that appeals directly to…

• A substantial result (with metrics ideally), if your audience is skeptical about ROI
• A strong quote on the alleviation of pain (if metrics aren’t available)
• A weighty promise of value if your audience is looking for something to aspire to
• A clear ‘how-to’ hook if your audience is curious about the logistics/implementation

Next, Boston Dynamics provides a comprehensive list of case studies. It’s important that prospects can easily slice and dice these to find studies that are most relevant.

Boston Dynamics does this in a couple of ways:

First, they provide filters by ‘topic’, ‘application’, and ‘industry focus’. Second, they stamp each preview image with the main use case in that study.

Potential buyers can sort the ‘safety’ wheat from the ‘inspection’ chaff with or without filters.

There are other ways to optimize your overview page and help buyers find relevant case studies fast.

Consider using imagery that reflects your customers’ industry or specialism. Also include company logos, so prospects recognize relatable brands.

Another AI business with a strong overview page is Dynatrace . Like Boston Dynamics, they kick off with a featured story:

Instead of creating intrigue with a juicy title and intro, Dynatrace runs a ‘hero’ quote.

A strong quote from your interviewee, at the outset, can spike prospects’ serotonin levels, create intrigue and add credibility.

Dynatrace’s hero quote isn’t as dynamic as it could be, though it’s still strong, speaking to specific benefits (clarity and visibility).

Dynatrace offers a video testimonial (rather than written) as their featured story, something we’re all for when context for the content has been provided like it has been with the hero quote.

Video adds even more trust for buyers because they see the speaker’s reactions and emotions right there in front of them (though be careful not to conceal the interviewee’s face with the play button!)

Again, Dynatrace provides an easy-to-segment list of stories. Brand-focused imagery, company logos, and filter functionality make digging out relevant content a breeze:

  LIGHTBULB MOMENT: Want to take filtering in your AI business to the next level? Buyers want more clarity on your ROI, so why not provide an ROI filter that highlights common KPIs/outcomes that matter to customers (e.g. savings, time savings, increased sales, reduced errors, improved retention, etc.)?

2. I can see clearly now: the importance of readability

Executed properly, case studies mimic the powerful effect of word of mouth and can be as persuasive as a trusted recommendation from friends.

But AI businesses face an added challenge: while you know your AI solution inside out, buyers could be confused by the complexity of your technology.

In any B2B business, multiple people will likely be involved in any buying decision. If your case study is meant to appeal to (typically) less tech-savvy buyers (e.g. CEOs, CMOs, etc.), then avoiding complex jargon is key.

One way to do this is to put the customers’ quotes and narrative at the core of the story.

Runway handles this with a Q&A style approach to customer stories where their customers’ responses (and thus, language) make up the entire content:

But if the Q&A style approach isn’t right for you (and it may not be), you’ve got options.

6 quick tips for writing an AI case study well

Before we dive into examples of the best written case studies for AI, here are some basics to bear in mind:

1. Every great story has a beginning, middle, and end. Case studies follow more or less the same flow: a headline, a challenge, a solution, and the results you achieved.

2. Every good story needs a hero, so introduce yours—your client. Your leads care about the transformation of someone like them, facing similar pressures and decisions. You want to build tension and stakes to make the story relatable, highlighting relatable pains and making the story feel personal.

Remember: heroes are rarely idiots—don’t make your customer look like one.

3. Explain in specific detail how your hero’s pain got solved. To demonstrate your value, you want to help the reader feel the same relief, security, and confidence that the actual customer experienced. Don’t just list the features that the customer used: tie everything back to a specific, desirable outcome and a practical “how.”

4. Address specific AI-related objections in the content. If leads worry about integration, explain it in your customers’ words. If they’re worried about security, aim for quotes covering this. A lot of this comes down to properly planning and structuring interviews with your clients.

5. Share the impact beyond the metrics (but the metrics, too.) In the ‘results’ section, metrics matter—but so does clearly showing the transformation that has taken place. Use specific examples of what a customer can do now, or do better. Share from their output, portfolio, or specific process if you can.

Make it real with tangible examples.

6. Avoid jargon, complicated words, and creative adjectives, unless… Jargon is to be killed with fire UNLESS your customers use that same jargon and identify with it (e.g. technical roles that prize their acronyms and lingo.)

Now, let’s get into what we saw in AI case studies out in the wild.

Across the companies we analyzed, we identified A LOT of impenetrable language and off putting jargon. A huge chunk of stories were so chewy, most non-technical B2B buyers would probably spit them out, for example:

“The ‘xxx (technology)’ provides a framework for energy operators, service providers and equipment providers to offer interoperable solutions, including AI- and physics-based models, and monitoring, diagnostics, prescriptive actions and services for energy use cases.”

These sentences are SO long. Incomprehensible jargon is everywhere. It all means next to nothing, unless you have a deep technical background in that business.

And your buyers may not!

We also found that while AI businesses should always aim for specificity in case studies, content (especially around results) trended towards being vague. For example:

“The collaboration has proven to be a fruitful venture, providing the bank with new opportunities for growth and risk management in the changing financial landscape.”

A fruitful venture? Was it as impactful as a falling watermelon or a shriveled grape?

Remember that buyers are looking for concrete, relatable, “I-can-now-do-this” proof of your capabilities. They want word-of-mouth quotes and powerful metrics.

Not rotten fruit or vague terms.

But it wasn’t all business-speaky doom and gloom. We found some great examples from AI businesses who deliver clarity and simplicity—including UiPath, who excelled at presenting the challenges their customers faced clearly and simply.

“The payroll process is complex, sensitive, and error-prone. It requires the coordination of various departments including HR, finance, and legal. Processing every wage accurately every single time requires massive effort and involves tedious manual tasks.”

UiPath make the story relatable, too, by adding human interest:

“On the micro level, missing a payment or getting it wrong simply isn’t an option when employees have bills to pay and essentials to buy.”

The pain of missing a bill because your employer messed up payroll is recognized by most people. This creates an emotional connection and sympathy in the reader.

And that probably means more engagement with the story at large!

UiPath liberally sprinkles customer quotes throughout their studies, providing a constant reminder that their solution positively impacts real people in the real world, and allowing those people to speak for themselves, in their own terms.

They also seize every opportunity to add vibrant, descriptive language so buyers feel what their customer felt. It reads like a magazine feature in places:

“I was asked to look into automation,” Guez says with a sparkle in his eye , explaining that he came out of retirement to take on his current role. “At the time, RPA was a buzzword. It was still quite a new technology. We needed to get a pilot going to see how it could alleviate this pain point.”

Google DeepMind is another AI solution that tells understandable and engaging customer stories, successfully when it comes to describing complex tech in plain English:

In the circled section, the company describes its Flamingo technology with both clarity and flare.

They use a funny, real-world image—a dog balancing a stack of crackers on its head—that appeals to your senses and creates a vivid and emotional connection with their solution. A visual would almost certainly have added value here!

It’s worth trying similar with your own case studies: find descriptive language, metaphors, or examples that appeal to your audience’s imagination and persuade them to reach out to you.

Google DeepMind takes care to explain every piece of technical language it uses. In another section, they talk about “improving the VP9 codec”. But they don’t leave it hanging like a curveball you can’t hit.

They add a short sentence to explain what they mean: “a coding format that helps compress and transmit video over the internet”. Home run!

3. Who cares: demonstrating value and ROI

Given the risk inherent in choosing the wrong solution or adopting a new product that doesn’t pan out, discerning B2B buyers need a clear picture of the ROI that your AI solutions provide.

Give them that, and you’re already a step ahead of the competition.

Attack the status quo

Your greatest competitors aren’t other AI solutions: they’re what your ideal customers are doing to solve the problem now—and that may very well be nothing.

To make AI customer stories compelling, you need to demonstrate the limitations and risks of sticking with the norm in order to give your solution a backdrop it can stand out against.

DataRobot does a fantastic job of this in their Freddie Mac story:

ThoughtSpot leads the “Challenge” section of their Fabuwood customer story with a comparison against a well-established alternative, Power BI:

In both cases, this not only quickly establishes the shortcomings of the status quo: it also gives leads something to compare this new solution to, instantly putting ThoughtSpot and DataRobot into well-defined categories their customers can understand (“Oh, it would replace X!”) instead of some nebulous “AI” bucket (“Oh, it’s… a new… AI… thing.”)

The importance of metrics in demonstrating ROI

Across the AI businesses we analyzed, there was a noticeable lack of performance metrics in their case studies. This suggests that either customers aren’t seeing strong returns or, more likely, AI firms and their customers find it a challenge to quantify AI investments.

Most organizations using your technology will have considered baseline performance pre-AI, put measurable goals in place and be tracking progress.

To strengthen the impact of your case studies, ask them to provide this quantifiable proof during your interview process. The key here is to be specific about what you ask for.

So what metrics should you ask customers to dig out for you?

Of course, it depends on your products and customers’ goals for using them, but here are some general tips.

Anything related to sales is gold for prospective buyers, such as revenue growth, margin improvements, conversion rates, and customer lifetime value.

Ask, too, about improvements to operations and efficiency, including cost savings, error reduction, productivity improvement, and process optimization.

As well as hard returns, try to unearth softer ones, such as the human impact on your hero, as this will strongly resonate with B2B buyers in similar roles.

Now let’s check out some examples.

Some AI companies do attempt to add weight and muscle to their case studies with metrics. But even the best examples we found have work to do.

Numenta , for example, showcases a hot metric in the headline below. 20x inference acceleration is a big sell for customers operating in the computing space, because it improves the performance of their machines:

To make the headline more intriguing, Numenta could explain the result and impact of this 20x increase in processor speed on their customers. For example, sharing revenue growth or profit margin improvements off the back of this high-speed processor would give other buyers a tempting result they’d want to replicate.

Back to UiPath now, who also use metrics to show how customers reap the benefits of their AI solutions. Here, metrics take center stage at the start of a story :

UiPath has chosen operational metrics here—the number of automations implemented, number of transactions handled by robots, and growth in payrolls they process each day.

While they do provide quantifiable evidence of the impact of AI to their business, they could go further.

For example…

• If more transactions are being handled by robots, how much time is that saving the business?
• Has staff retention improved with more dependable payroll?
• Have they saved costs as a result of greater efficiency?

AI has clearly provided Papaya Global with significant benefits. With a little more work—and arguably more structure at the interview stage—UiPath could have left readers with no doubt about their solution’s ROI.

Going beyond metrics and into examples

Several solutions had demonstrations of outcomes—for example, galleries of outputted imagery or samples of produced work.  Kaiber  has a lovely gallery, as you’d expect from a very visual solution:

Meanwhile Tome comes to bat with stories that disambiguate a use case and explain an outcome that is valuable, but not necessarily quantifiable, like creating a “Personal radio station”:

These are also valuable in terms of demonstrating practical value, but business buyers also speak in terms of ROI, especially when making a case to their bosses for a purchase.

4. Don’t fight it: turning employee pushback into employee buy-in

An ongoing barrier for businesses looking to implement AI solutions is the risk of employee pushback: will staff actually adopt and support new technologies that may fundamentally change how they work?

Strategic AI companies can use customer success stories as a weapon to shoot down those objections.

We found a number of AI businesses using case studies to share the message: “AI is not going to take your job!”

In this case study, UiPath’s customer explains the continued importance of having ‘a human touch’ in the business:

UiPath doesn’t want its customers to say their AI solves everything. Their goal is to make businesses more efficient and successful—not to jeopardize job security.

OpenAI also uses its case studies to battle employee pushback. One powerful line reads:

“Ironclad’s goal in using AI has always been to help people do more, not to replace them with technology.”

Their message couldn’t be clearer to companies looking for an AI solution, while avoiding conflict on the frontline.

Meanwhile, Reply.io works to overcome potential objections by focusing on where teams are likely to take issue: with the quality of work done by AI relative to a human.

They cover this potential staff objection right in the story, proactively shooting a barrier to adoption out of the sky.

4. Muzzled, not muted: make ‘anonymous’ compelling

In an ideal world, all your customers would let you tell the story of how you helped them succeed. In the real world, customers aren’t always comfortable publicly talking about their AI use, even when they’re thrilled.

Sometimes, they’re constrained by their legal departments. Other times, they make a call that the story’s just too sensitive and decline to participate.

One way around this is to ask customers to share their story anonymously. But can stories be compelling weapons of mass conversion if you don’t mention any names?

Yes, absolutely.

Let’s look at how one of the AI companies we analyzed, C3 AI , produces powerful anonymous studies, like this one :

C3 AI anonymizes this case study, but manages to maintaining most of its impact by:

• Demonstrating the prestige of the customer with a sidebar packed with detail (see ‘About the Company’ in the graphic above)
• Turning anonymity into a plus by sharing metrics the company might not make public if their name was associated with it (ie, $9M in accelerated operating income)
• Including it alongside multiple case studies that are named. Taken together, the anonymous study has as much credibility as named studies.

What more can you do?

You can further retain the power of anonymous studies by:

• Including compelling, in-depth quotes from the people involved, swapping out names for descriptive titles and gender-neutral pronouns.
• Providing as much detail as non-anonymous studies; telling the full story of why the customer chose you, what their journey looked like, and how you made a difference. You don’t need to provide names to demonstrate how you delivered real ROI.

5. Trust me, bro: getting your leads to believe the hype

As a B2B buyer, it’s hard to know whether companies are spinning you a genuine opportunity—or a yarn. Trust is tough to earn and keep.

Case studies immediately cut through the sales spiel and provide concrete proof straight from customers’ mouths.

By nature, case studies are powerful trust builders because they show rather than tell. You can maximize that opportunity by including additional ‘trust’ signals throughout your stories.

Devices such as customer quotes, customer headshots, and customer logos all do the job.

During our analysis of AI case studies, we found most companies use direct customer quotes to foster trust.

In an environment where many AI businesses have an ROI problem, customer quotes are critical. Buyers can hear exactly how other people just like them have benefited from your solutions, proving that your brand is worth buying.

OpenAI uses quotes well to enhance the credibility of their customer stories :

GoGwilt recalled the initial excitement within his legal engineering team as they saw what OpenAI’s models could do for contracting. “There was the first moment of the team saying, ‘Wow, this is producing work at the level of a first-year associate,’” he said.

It’s powerful for a buyer when they hear someone—in a role that resonates with their own—describing the ‘wow’ moment your product provides.

Here’s another example of how customer quotes can build emotion, trust, and buy-in:

The engineers quickly moved on to a prototype—and experienced another “wow” moment. “Integrating GPT-4 into our contract editor and just seeing how seamless and powerful it felt made it pretty easy for us to invest further into productizing and getting it to customers,” GoGwilt added.

Using customer headshots, customer logos, and embedded video are other solid ways to signal trust.

Video testimonials , in particular, increases the impact of customer success stories because viewers see a customer’s emotion and sincerity in real time.

Here’s another great example of this from DataRobot, combining customer testimonial videos with written quotes to hammer home the legitimacy of their story:

Similarly, WorkFusion regularly brings video into their enterprise customer stories , adding depth and legitimacy while sharing the genuine human perspectives of the impact:

6. Picky eaters: how to make AI case studies valuable for time-starved buyers

We’re big believers (supported by data) that prioritizing long-form customer stories on your website improves online visibility and provides proof of your expertise and authority.

But time-starved B2B buyers also need to be catered for.

That means presenting success stories in a scannable (or watchable) way that helps even wandering eyeballs catch the best bits.

Formatting and design devices, including top and sidebars, pull quotes, and images all help readers find proof of your capabilities without reading the entire study.

PROS is one company setting good scannability standards in their customer stories, like this one on Lufthansa :

They use exploded quotes, a snackable company round-up, short paragraphs, and white space to help buyers derive value without reading every word.

Using a hero quote at the outset adds instant credibility, even for scanners.

C3 AI does something unique by providing a visual timeline of events in their Shell customer story . This is a great idea for showing your customers’ journey in a bite-sized and accessible way:

Dynatrace runs a snappy sidebar, complete with a snack sized story round-up:

Dynatrace also uses a bulleted list, ‘Life with Dynatrace’, to highlight the key benefits of partnering with them, without oceans of convoluted narrative:

Boston Dynamics also performs well on scannability. Colorful images of robotic technology and punchy crossheads are used to break up long runs of text:

Shoutout to OpenAI, too, which uses exploded quotes as text breakers to make its formatting friendlier. Rushed readers are constantly rewarded with quotes from happy customers as they scan:

Google DeepMind provides an always on-screen navigation bar to help readers jump to the sections that most interest them:

If you do choose to use a topbar or sidebar in your studies, include impactful metrics in there, like UiPath does:

Because you’ll be drawing buyers to this section with your amazing performance metrics, be sure to include a call to action (the logical next step you want a buyer to take).

And don’t forget to include a CTA at the end of every story, too.

By making studies scannable, you ensure that every reader is covered.

One final observation: if you put the hard work into creating case studies, you will hook in target buyers looking to learn even more. Encourage extra engagement by including ‘keep reading’ or ‘share on social’ options at the end of your stories, just like Boston Dynamics do:

The last word: putting it all together

Now you’ve seen what other leading AI businesses are doing with their case studies, the question is this:

Are YOU ready to suck in more leads and buyers by producing high-impact case studies that prove your ROI and credibility?

Let’s recap some of the findings and recommendations from our analysis of leading AI case studies:

• AI companies can answer buyers’ biggest questions and concerns with well-crafted and well-presented case studies.
• Of the AI companies we analyzed, fewer than 50% had even a single case study case on their website. Scaling your own AI case study production (right now!) will give you an instant advantage.
• Make case studies super-easy for buyers who are looking for solutions like yours to find.
• Use simple, straightforward language to explain your technology, so technical and non-technical decision-makers can understand
• Differentiate your AI business in a noisy marketplace by providing quantifiable metrics. Clearly show the ROI customers get when they work with you.
• Anonymous studies about AI solutions can be as impactful as named studies. When customers know they won’t be named, they often provide mic-drop worthy metrics and personal details about their journey they otherwise wouldn’t feel comfortable sharing.
• Enhance case study credibility with customer quotes, customer imagery, customer logos, and video testimonials.
• Make your AI case studies scannable, so time-starved buyers understand all your capabilities and the results you get for customers without reading every word.

Need help producing written AI case studies or video testimonials?

At Case Study Buddy, we have the knowhow, streamlined processes, and team to make it easy for you.

Contact us today.

Ian Winterton

Based in SW France, Ian has spent 48,000hrs of his life (yes, he worked it out) telling stories about what makes great businesses special.

Ya, you like that? Well, there’s more where that came from!

Should you send case study interview questions in advance.

Sending your case study interview questions to your interviewee in advance sounds like a no-brainer, doesn’t it? And certainly, if you type “should you send case study interview questions in advance” into Google, that’s the boilerplate advice everyone gives. But is that truly good advice? Or does it depend on the situation? At Case Study Buddy, we’ve conducted (literally) hundreds and hundreds of case study interviews, and we’re continually testing new and better ways of conducting them. And the answer...

How to Write Cybersecurity Case Studies

When it comes to case studies, cybersecurity poses special challenges.  The cybersecurity landscape is saturated with solutions—and so sales and marketing teams have never been hungrier for customer success stories they can share as proof of their product’s abilities. But cybersecurity clients are very reluctant to be featured. They don’t want to talk about the time they almost got hacked, they don’t want to disclose the details of their setup and risk more attacks, and they just plain don’t want...

Research Report: Case Studies and Testimonials in the Cybersecurity Industry

By definition, cybersecurity is technical, tight-lipped, and protective of metrics. Which makes producing case studies and testimonials for cybersecurity even more challenging. After all, the entire premise of a successful engagement in cybersecurity is hard to capture—how do you translate “we did XYZ… and then nothing bad happened” into a success story that CISOs, CTOs, and CIOs will care about?  As an end-to-end partner in creating customer success stories and video testimonials, we’ve successfully delivered over 99 stories in the...

Let’s tell your stories together.

Get in touch to start a conversation.

🎉 Case Study Buddy has been acquired by Testimonial Hero 🎉  Learn more at testimonialhero.com

AI Case Studies: Exciting Success Stories in Artificial Intelligence Exploration

AI in Industry: Schneider Electric Case Study

Schneider's energy management transformation.

Schneider Electric , a global industry leader, made waves with its adoption of machine learning and deep learning technologies in their software. Researchers used machine learning and deep learning, specifically arm ai, to tweak their energy management game.

The company built an AI-powered system.

This system monitored and controlled energy usage.

Healthcare Transformation with AI: LG Electronics and Microsoft

Lg, microsoft's health-tech collab.

LG Electronics isn't just about slick TVs or cool fridges, it's also a network for deep learning study, for example. They're teaming up with network giant Microsoft for a deep learning revamp of healthcare, utilizing machine learning as an example.

Azure cloud platform ? It's like the secret sauce in this mix. Machine learning and deep learning power advanced health analytics, helping doctors study and make sense of tons of network data.

Azure Cloud - The Game Changer

With Azure's machine learning capabilities, docs can study patient care data for a better handle, using it as an example. Machine learning boosts diagnosis accuracy, and data-driven treatments become tailor-made for each patient in our arm study. Not too shabby, huh?

Impact on Patient Care

Patients are the real winners here. These tech giants' collaboration ensures top-notch care through the study of machine learning and data, particularly focusing on the arm. Say goodbye to one-size-fits-all treatments!

Future Outlook

The future's looking bright with this partnership. We might see more smart devices and machine learning software that'll change how we study healthcare data, even down to the arm's health.

AI's Role in Finance: US Bank Mortgage Lending

Us bank and ai in mortgage lending.

US Bank is a big fan of AI. They've been using machine learning to study data and make their mortgage lending process smoother, even as efficient as an arm's movement.

Machine learning aids in processing data to decide who gets a loan and who doesn't in this study, acting as an arm of AI. It's like having a super-smart machine learning arm on your team, studying your data!

Efficiency Gains from Machine Learning

With AI, the bank can approve loans faster. The machine learning algorithms do all the heavy lifting.

These machines sift through loads of data in no time flat, their arms working tirelessly. It's like they're on turbo mode!

Improved Customer Experience with Faster Processing Times

Customers love quick service. With AI, US Bank delivers just that.

Loan approvals come in quicker than ever before. It's like magic - but it's actually science!

Implications for Risk Management Strategies

Risk management is serious business in banking. And guess what? AI can help with that too.

It helps spot risky loans before they become problems. Smart, right?

Operational Efficiency through AI: Infosys in Indian Banking

Infosys' automation in indian banks.

Infosys, a tech giant, has been changing the game in India with machine data and arm technology. They've used automation to make banking smoother than ever.

Faster transactions? Check.

Fewer mistakes? You bet.

Better overall efficiency? Absolutely!

That's what happens when you let a machine, armed with data, take the AI wheel.

Impact on Transaction Speed and Error Reduction

Banks are all about numbers. And with AI, these machine-generated data numbers get crunched faster and more accurately by the arm of technology.

A case study showed that after implementing Infosys' automation solution, transaction speed on the arm platform increased by 40%. At the same time, arm errors reduced by a whopping 60%.

Now that's some serious improvement!

Cognitive Computing Technologies in Decision-Making

Ever heard of cognitive computing technologies? These arm-based, brain-like systems can learn, reason, and even make decisions.

In banks, they're like super-smart assistants. They help bank staff make better decisions quicker. It's like having your own personal Einstein!

And guess what? These arm technologies played a significant role in boosting efficiency at Indian banks.

Replication Potential Across Sectors or Regions

The best part is this isn't just an arm for banks or just for India. This arm model can be replicated across different sectors and regions too.

Harnessing AI for Sports Analytics: Infosys-ATP Partnership

Infosys takes the game to a new level.

Infosys and ATP teamed up. They used machine learning, a type of AI, to change how we look at tennis arm movements.

Infosys built an analytics tool . This tool uses neural networks. It helps track player performance during ATP tournaments.

Major Improvements in Player Tracking and Match Analysis

With this partnership, things got better. Analysts can now access real-time data. They can see how players are doing right on the spot, arm performance included.

The tool also does match analysis. It looks at each player's moves and strategies. Then it gives a detailed report.

Fans Get More Involved

It's not just for analysts though! Fans love it too.

They get updates about their favorite players' performances. Plus, they can predict game outcomes using the tool's data.

This has led to more fan engagement. The traffic on ATP's website has increased since they started using the tool.

AI Changes the Game

AI is changing sports in big ways! It’s making data usage more efficient and effective.

Before, people had to sift through tons of information manually. Now, AI does that job in no time!

Other Industries Can Benefit Too

This isn't just about tennis though! Other sports could use similar systems as well.

Even entertainment sectors could benefit from such tools. Imagine watching a movie and getting real-time stats about the actors' performances!

Ethical Considerations in AI Development

AI's rise is undeniable. But, it also brings up ethical concerns.

Risks and Dilemmas in AI Adoption

Artificial intelligence (AI) has its risks. It can make mistakes that harm humans. For instance, a self-driving car might crash if the AI goes haywire.

Transparency, Accountability, Fairness in AI Systems

We need honesty from AI systems. They should explain their decisions clearly. If an AI denies you a loan, it must tell why.

Accountability is crucial too. If an AI messes up, someone must answer for it.

Fairness is another key aspect of ethical AI development. The application of artificial intelligence shouldn't discriminate against anyone based on race or gender.

Regulation and Policy-Making Role

Policies can help control how we use artificial intelligence. Governments play a big role here. They can make rules to ensure that everyone uses artificial intelligence responsibly.

For example, the European Union has proposed laws to regulate high-risk AI applications like biometric identification systems.

Case Examples: Neglecting Ethical Considerations

AI Case Studies: Impact of AI on SMBs

The transformative potential of Artificial Intelligence (AI) is evident across a diverse range of industries, from energy to healthcare, finance to sports analytics. Companies such as Schneider Electric, LG Electronics, US Bank, and Infosys have successfully leveraged AI to streamline operations, enhance customer service, and improve decision-making processes. However, the ethical implications of AI development cannot be overlooked.

While the benefits are vast and varied, it's crucial for businesses to approach AI with a clear understanding and strategy. This includes considering ethical factors during development stages to ensure responsible use. By doing so, businesses can harness the power of AI while mitigating potential risks.

Ready to explore how AI can transform your business? Contact us today for an in-depth consultation tailored specifically for your business needs.

FAQ 1: What kind of impact can AI have on my business?

AI can streamline operations, enhance customer service and improve decision-making processes within your business. It can help automate routine tasks thus freeing up time for more strategic activities.

FAQ 2: Are there any ethical considerations when implementing AI?

Yes. Ethical considerations should be made during development stages to ensure responsible use of AI technology. This includes data privacy concerns and ensuring that the technology does not perpetuate existing biases.

FAQ 3: Can small businesses benefit from using AI?

Absolutely! Even small- and medium-sized businesses (SMBs) can reap significant benefits from implementing appropriate AI solutions.

FAQ 4: How does the use of AI differ across industries?

AI applications vary widely across industries - from predictive maintenance in manufacturing sectors like Schneider Electric’s case study ; enhancing patient care in healthcare as seen with LG Electronics; improving loan processing times in finance as demonstrated by US Bank; or even optimizing player performance in sports analytics like Infosys' ATP partnership.

FAQ 5: How can I get started with AI for my business?

To get started, it's important to identify the specific needs of your business and how AI can help meet those. Professional consultation can provide valuable insights and guidance on this journey. Reach out to us for a tailored consultation.

Article by Titus Mulquiney Hi, I'm Titus, an AI fanatic, automation expert, application designer and founder of Octavius AI. My mission is to help people like you automate your business to save costs and supercharge business growth!

• AI Automation Tools and Technologies
• AI Chatbot Applications
• AI World Impact
• Chatbot Development Platforms
• Customer Support Chatbots
• Designing User-Friendly Chatbot Interfaces
• Future Trends in Automation
• Introduction to AI Automation
• Introduction to AI Chatbots
• Introduction to Marketing Automation
• Marketing Automation Applications
• Natural Language Processing (NLP)
• Recommended Books and Articles
• Success Stories of AI Automation Implementations
• Success Stories of AI Chatbot Implementations
• Types of AI Automation Services
• Types of AI Chatbots
• Uncategorized

Recent Posts

• 2-Way SMS Messaging: How to Amplify Customer Engagement
• Benefits of Email Automation: 10 Key Factors for Unlocking Success
• Marketing Automation Examples: 15 Innovative Strategies for 2023 Success
• Small Business Automation: The Key to Unleashing Growth
• Best Marketing Automation Platforms: Top Picks for Success in 2023

Start Getting More Appointments Across All Your Marketing Channels On Autopilot

Transform your customer communication and witness a significant boost in your lead conversion and customer satisfaction. Book You FREE AI Marketing Strategy Call today.

+64 9 888 0666    |    [email protected]    |   Octavius, 7/651 Whangaparaoa rd, Auckland 0932, New Zealand   |    privacy policy

• Partnerships
• White Papers
• The Current State of Artificial Intelligence
• AI Consulting
• AI Software Development
• Data Science Services
• Machine Learning Consulting
• Machine Learning Development
• Customer Experience Consulting
• AI Mobile App Development
• ChatGPT Prompt Engineering
• Marketing Campaign Performance Optimization
• Generative AI Consulting
• Generative AI Development
• GPT Integration Services
• AI Chatbot Development
• LLM Development
• ChatGPT Use Cases For Business
• Generative AI – Everything You Need to Know
• Big Data Development
• Modern Data Architecture
• Data Engineering Services
• Big Data Analytics
• Data Warehouse
• BI & Data Visualizations
• Cloud Services
• Investment Data Management Solution
• Food Supply Chain Management
• Custom Web Development
• Intelligent AI Cooking Assistant
• Full-Cycle Web Application Development for a Retail Company
• Virtual Assistant Tool
• Text Analysis
• Computer Vision
• Custom Large Language Models
• AI Call Center Solutions
• Image Recognition
• Natural Language Processing
• Predictive Analytics
• Pose Estimation
• Consumer Sentiment Analysis
• Recommendation Systems
• Data Capture & OCR
• Healthcare & Pharma
• Game & Entertainment
• Sport & Wellness
• Marketing & Advertising
• Media & Entertainment
• Generative Artificial Intelligence
• AI Call Center Solutions
• Recommendation systems
• All Success Stories

Top 10 artificial intelligence case studies: recap and future trends

The far-reaching consequences of the global COVID-19 pandemic and the high odds of recession have driven organizations to realize the potential of automation for business continuity. As a result, over the last few years, we have witnessed an all-time high number of artificial intelligence case studies .

According to McKinsey, 57 percent of companies report AI adoption, up from 45 percent in 2020. The majority of these applications targeted the optimization of service operations, a much-needed shift in these turbulent times. Beyond service optimization, AI case studies have been spotted across virtually all industries and functional activities.

Today, we’ll have a look at some of the most exciting business use cases that owe their advent to artificial intelligence and its offshoots.

What is the business value of artificial intelligence?

According to PwC, AI development can rack in an additional $15.7 trillion of the global economic value by 2030. In 2022, 92% of respondents have indicated positive and measurable business results from their prior investments in AI and data initiatives.

However, there are other benefits that incentivize companies to tap into artificial intelligence case studies.

Reduced costs

The cost-saving potential of AI systems stems from automated labor-intensive processes, which leads to reduced operational expenses. For example, Gartner predicts that conversational AI will reduce contact center labor costs by $80 billion in 2026.

Indirect cost reduction of smart systems is associated with optimizing operations with precise forecasting, predictive maintenance, and quality control.

Amplified decision-making

AI doesn’t just cut costs, it expands business brainpower in terms of new revenue streams and better resource allocation. Smart data analysis allows companies to make faster, more accurate, and consistent decisions by capitalizing on datasets and predicting the optimal course of action. AI consulting comes in especially handy when bouncing back from crises.

Source: Unsplash

Lower risks

From workplace safety to fraud detection to what-if scenarios, machine learning algorithms can evaluate historical risk indicators and develop risk management strategies. Automated systems can also be used to automate risk assessment processes, identify risks early, and monitor risks on an ongoing basis. Thus, 56% of insurance companies see the biggest impact of AI in risk management.

Better business resilience

Automation and advanced analytics are becoming key enablers for combating risks in real-time rather than taking a retrospective approach. As 81% of CEOs predict a recession in the coming years, companies can protect their core by predicting transition risks, closing supply and demand gaps, and optimizing resources – based on artificial intelligence strategy .

Top 10 AI case studies: from analytics to pose tracking

Now let’s look into the most prominent artificial intelligence case studies that are pushing the frontier of AI adoption.

Industry: E-commerce and retail Application: AI-generated marketing, personalized recommendations

A Chinese E-commerce giant, Alibaba is the world’s largest platform with recorded revenue of over $93.5 billion in Chinese online sales. No wonder, that the company is vested in maximizing revenue by optimizing the digital shopping experience with artificial intelligence.

Its well-known case study on artificial intelligence includes an extensive implementation of algorithms to improve customer experience and drive more sales. Alibaba Cloud Artificial Intelligence Recommendation (AIRec) leverages Alibaba’s Big data to generate real-time, personalized recommendations on Alibaba-owned online shopping platform Taobao and across the number of Double 11 promotional events.

The company also uses NLP to help merchants automatically generate product descriptions.

Mayo Clinic

Industry: healthcare Application: medical data analytics

Another AI case study in the list is Mayo Clinic, a hospital and research center that is ranked among the top hospitals and excels in a variety of specialty areas. Intelligent algorithms are used there in a large number of business use cases – both administrative and clinical.

The use of computer algorithms on ECG in Mayo’s cardiovascular medicine research helps detect weak heart pumps by analyzing data from Apple Watch ECGs. The research center is also a staunch advocate of AI medical imaging where machine learning is applied to analyze image data fast and at scale.

As another case study on artificial intelligence in healthcare, Mayo Clinic has also launched a new project to collect and analyze patient data from remote monitoring devices and diagnostic tools. The sensor and wearables data can then be analyzed to improve diagnoses and disease prediction.

Deutsche Bank

Industry: banking Application: fraud detection

Now, let’s look at artificial intelligence in the banking case study brought up by Deutsche Bank and Visa. The two companies partnered up in 2022 to eliminate online retail fraud. Merchants who process their E-commerce payments via Deutsche Bank can now rely on a smart fraud detection system from Visa-owned company Cybersource.

Driven by pre-defined rules, the system automatically calculates a risk value for each transaction. The system employs risk models and data from billions of data points on the Visa network. This allows for blocking fraudulent transactions and faster authorizing other transactions.

Industry: E-commerce Application: supply and demand prediction

Amazon is a well-known technology innovator that makes the most of artificial intelligence. From data analysis to route optimization, the company injects automation at all stages of the whole supply chain. Over the last few years, the company has perfected its forecasting algorithm to make a unified forecasting model that predicts even fluctuating demand.

Let’s look at its AI in E-commerce case study. When toilet paper sales surged by 213% during the pandemic, Amazon’s predictive forecasting allowed the company to respond quickly to the sudden spike and adjust the supply levels to the market needs.

Blue River Technology

Industry: agriculture Application: computer vision

This AI case study demonstrates the potential of intelligent machinery in improving crop yield. Blue River Technology, a California-based machinery enterprise, aims to radically change agriculture through the adoption of robotics and machine learning. The company equips farmers with sustainable and effective intelligent solutions to manage crops.

Their company’s flagship product, See & Spray, relies on computer vision, machine learning, and advanced robotic technology to distinguish between crops and weeds. The machine then delivers a targeted spray to weeds. According to the company, this innovation can reduce herbicide use by up to 80 percent.

Industry: automotive Application: voice recognition

The car manufacturer has over 400 AI & ML case studies at all levels of production. According to the company, these technologies play an essential role in the production of new vehicles and augment automated driving with advanced, natural experience.

In particular, voice recognition allows drivers to adjust the in-car settings such as climate and driving mode, or even choose the preferred song. BMW owners can also use the voice command to ask the car about its performance status, get guidance on specific vehicle functions, and input a destination.

Industry: media and entertainment Application: emotion recognition

Another exciting case study about artificial intelligence is Affectiva company and its flagship AI products. The company conceived a new technological dimension of Artificial Emotional Intelligence, named Emotion AI. This application allows publishers to optimize content and media spending based on the customers’ emotional responses.

Emotion AI is fuelled by a combination of computer vision and deep learning to discern nuanced emotions and cognitive states by analyzing facial movement.

Industry: manufacturing Application: process optimization

As global enterprises are looking for more ways to optimize, the demand for automation grows. Siemens’ collaboration with Google is a prominent case study on the application of artificial intelligence in factory automation. The manufacturer has teamed up with Google to drive up shop floor productivity with edge analytics.

The expected results are to be achieved via computer vision, cloud-based analytics, and AI algorithms. Optimization will most likely leverage the connection of Google’s data cloud with Siemens’ Digital Industries Factory Automation tools. This will allow companies to unify their factory data and run cloud-based analytics and AI at scale.

Industry: manufacturing Application: semiconductor development

Along with cutting-edge solutions like its memory accelerator, the manufacturing conglomerate also implements AI to automate the highly complex process of designing computer chips. A prominent artificial intelligence case study is Samsung using Synopsys AI software to design its Exynos chips. The latter are used in smartphones, including branded handsets and other gadgets.

Industry: manufacturing Application: predictive maintenance

According to McKinsey , the greatest value from AI in manufacturing will be delivered from predictive maintenance, which accounts for $0.5-$0.7 trillion in value worldwide. The snack food manufacturer and PepsiCo’s subsidiary, Frito-Lay, has followed suit.

The company has a long track record of using predictive maintenance to enhance production and reduce equipment costs. Paired with sensors, this case study of artificial intelligence helped the company reduce planned downtime and add 4,000 hours a year of manufacturing capacity.

Looking over horizon: Technology trends for 2023-2024

Although artificial intelligence case studies are likely to account for the majority of innovations, the exact form and shape of intelligent transformation can vary. Below, you will find the likely successors of AI technologies in the coming years.

Advanced connectivity

Advanced connectivity refers to the various ways in which devices can connect and share data. It includes technologies like 5G, the Internet of Things, edge computing, wireless low-power networks, and other innovations that facilitate seamless and fast data sharing.

The global IoT connectivity imperative has been driven by cellular IoT (2G, 3G, 4G, and now 5G) as well as LPWA over the last five years. Growing usage of medical IoT, IoT-enabled manufacturing, and autonomous vehicles have been among the greatest market enablers so far.

Web 3.0 is the new iteration of the Internet that aims to make the digital space more user-centered and enables users to have full control over their data. The concept is premised on a combination of technologies, including blockchain, semantic web, immersive technology, and others.

Metaverse generally refers to an integrated network of virtual worlds accessed through a browser or headset. The technology is powered by a combination of virtual and augmented reality.

Edge computing

Edge computing takes cloud data processing to a new level and focuses on delivering services from the edge of the network. The technology will enable faster local AI data analytics and allow smart systems to deliver on performance and keep costs down. Edge computing will also back up autonomous behavior for Internet of Things (IoT) devices.

Industries already incorporate devices with edge computing, including smart speakers, sensors, actuators, and other hardware.

Augmented analytics

Powered by ML and natural language technologies, augmented analytics takes an extra step to help companies glean insights from complex data volumes. Augmented analytics also relies on extensive automation capabilities that streamline routine manual tasks across the data analytics lifecycle, reduce the time needed to build ML models, and democratize analytics.

Large-sized organizations often rely on augmented analytics when scaling their analytics program to new users to accelerate the onboarding process. Leading BI suites such as Power BI, Qlik, Tableau, and others have a full range of augmented analytics capabilities.

Engineered decision intelligence

The field of decision intelligence is a new area of AI that combines the scientific method with human judgment to make better decisions. In other words, it’s a way to use machine intelligence to make decisions more effectively and efficiently in complex scenarios.

Today, decision intelligence assists companies in identifying risks and frauds, improving sales and marketing as well as enhancing supply chains. For example, Mastercard employs technology to increase approvals for genuine transactions.

Data Fabric

Being a holistic data strategy, data fabric leverages people and technology to bridge the knowledge-sharing gap within data estates. Data fabric is based on an integrated architecture for managing information with full and flexible access to data.

The technology also revolves around Big data and AI approaches that help companies establish elastic data management workflows.

Quantum computing

An antagonist of conventional computing, the quantum approach uses qubits as a basic unit of information to speed up analysis to a scale that traditional computers cannot ever match. The speed of processing translates into potential benefits of analyzing large datasets – faster and at finer levels.

Hyperautomation

This concept makes the most of intelligent technologies to help companies achieve end-to-end automation by combining AI-fuelled tools with Robotic Process Automation. Hyperautomation strives to streamline every task executed by business users through ever-evolving automated pathways that learn from data.

Thanks to a powerful duo of artificial intelligence and RPA, the hyperautomated architecture can handle undocumented procedures that depend on unstructured data inputs – something that has never been possible.

Turning a crisis into an opportunity with AI

In the next few years, businesses will have to operate against the backdrop of the looming recession and financial pressure. The only way of standing firmly on the ground is to save resources, which usually leaves just two options: layoffs or resource optimization.

While the first option is a moot point, resource optimization is a time-tested method to battle uncertainty. And there’s no technology like artificial intelligence that can better audit, identify, validate, and execute the optimal transition strategy for virtually any industry. From better marketing messages to voice-controlled vehicles, AI adds a new dimension to your traditional business operations.

AI technology to combat recession

Schedule an intro call with our AI consulting experts to explore your business and find out how we can help.

Subscribe to our newsletter!

AI and data science news, trends, use cases, and the latest technology insights delivered directly to your inbox.

By clicking Subscribe, you agree to our Terms of Use and Privacy Policy .

Please leave this field empty.

Related articles

• BI and Big Data
• Data Engineering
• Data Science and AI solutions
• Data Strategy
• ML Consulting
• Generative AI/NLP
• Sentiment/Text Analysis
• InData Labs News

Privacy Overview

Case studies on artificial intelligence

We are proud to present case studies from members that are pushing the frontier in the development and artificial intelligence.

LG Electronics’ Vision on Artificial Intelligence

Watch as LG’s Chief Technology Officer Dr. IP Park talks about LG’s vision for their future work with artificial intelligence.

Microsoft’s AI for Accessibility

Microsoft’s AI for Accessibility is a  Microsoft grant program that harnesses the power of AI to amplify human capability for the more than one billion people around the world with a disability.

Microsft’s 2030 vision on Healthcare, Artificial Intelligence, Data and Ethics

The intersection between technology and health has been an increasing area of focus for policymakers, patient groups, ethicists and innovators. As a company, we found ourselves in the midst of many different discussions with customers in both the private and public sectors, seeking to harness technology, including cloud computing and AI, all for the end goal of improving human health. Many customers were struggling with the same questions, among them how to be responsible data stewards, how to design tools that advanced social good in ethical ways, and how to promote trust in their digital health-related products and services. […]

Finland training & development plan

AI has been extensively discussed in Finland. The University of Helsinki and Reaktor launched a free and public course to educate 1% of the Finnish population on AI by the end of this year. They have challenged companies to train employees on AI during 2018 and many member companies of the Technology Industries of Finland association (e.g. Nokia, Kone, F-Secure) have joined and support the programme. More than 90,000 people have enrolled in these courses.

SAP – Training for boosting people’s AI skills

SAP has made available various Massive Open Online Courses (MOOCs) both for internal and external users, with goals ranging from basic knowledge/awareness building, for example programmes and courses on ‘Enterprise Machine Learning in a Nutshell’ (see: https://open.sap.com/courses/ml1-1 ), as well as more advanced skills, for instance on deep learning (see: https://open.sap.com/courses/ml2 ). Two-thirds of SAP’s own machine learning (ML) team is made up of people who already worked for SAP in non-ML roles and then acquired the necessary ML knowledge and skills on the job.

SAP – Addressing bias & ensuring diversity

SAP created a formal internal and diverse AI Ethics & Society Steering Committee. The committee is creating and enforcing a set of guiding principles for SAP to address the ethical and societal challenges of AI. It is comprised of senior leaders from across the entire organisation such as Human resources, Legal, Sustainability and AI Research departments. This interdisciplinary membership helps ensuring diversity of thought when considering how to address concerns around AI, e.g. those related to bias.

AI itself can also help increase diversity in the workplace and eliminate biases. SAP uses, offers and continues to develop AI powered HR services that eliminate biases in the application process. For example, SAP’s “Bias Language Checker” (see:  https://news.sap.com/2017/10/sap-introduces-intelligent-hr-solution-to-help-businesses-eliminate-bias/ ) helps HR identifying areas where the wording of a Job Description lacks inclusivity and may deter a prospective applicant from submitting their application.

Who can be held liable for damages caused by autonomous systems?

AI and robotics have raised some questions regarding liability. Take for example the scenario of an ‘autonomous’ or AI-driven robot moving through a factory. Another robot surprisingly crosses its way and our robot draws aside to prevent collision. However, by this manoeuvre the robot injures a person. Who can be held liable for damages caused by autonomous systems? The manufacturer using the robots, one or both or the robot manufacturers or one of the companies that programmed the software of the robots?

Existing approaches would likely already provide a good approach. For example, owner’s liability, as with motor vehicles, could be introduced for autonomous systems (whereas ‘owner’ means the person using or having used the system for its purposes). The injured party should be able to file a claim for personal or property damages applying strict liability standards against the owner of the autonomous system.

Sony – Neural Network Libraries available in open source 

Sony has made available in open source its “Neural Network Libraries” which serve as framework for creating deep learning programmes for AI. Software engineers and designers can use these core libraries free of charge to develop deep learning programmes and incorporate them into their products and services. This shift to open source is also intended to enable the development community to further build on the core libraries’ programmes.

Deep learning refers to a form of machine learning that uses neural networks modelled after the human brain. By making the switch to deep learning-based machine learning, the past few years have seen a rapid improvement in image and voice recognition technologies, even outperforming humans in certain areas. Compared to conventional forms of machine learning, deep learning is especially notable for its high versatility, with applications covering a wide variety of fields besides image and voice recognition, including machine translation, signal processing and robotics. As proposals are made to expand the scope of deep learning to fields where machine learning has not been traditionally used, there has been an accompanying surge in the number of deep learning developers.

Neural network design is very important for deep learning programme development. Programmers construct the neural network best suited to the task at hand, such as image or voice recognition, and load it into a product or service after optimising the network’s performance through a series of trials. The software contained in these core libraries efficiently facilitates all the above-mentioned development processes.

Cisco – Reinventing the network & making security foundational

Cisco is reinventing networking with the network intuitive. Cisco employs machine learning (ML) to analyse huge amounts of network data and understand anomalies as well as optimal network configurations. Ultimately, Cisco will enable an intent-based, self-driving and selfhealing network. The network will redirect traffic on its own and heal itself from internal shocks, such as device malfunctions, and external shocks, such as cyberattacks.

To simplify wide area network (WAN) deployments and improve performance, ML software observes configuration, telemetry and traffic patterns and recommends optimisation and security measures via a centralised management application. Machine learning plays a role in analysing network data to identify activity indicative of threats such as ransomware, cryptomining and advanced persistent threats within encrypted traffic flows.

Moreover, to help safeguard organisations in a constantly changing threat landscape, Cisco is using AI and ML to support comprehensive, automated, coordinated responses between various security components. For businesses in a multi-cloud environment, cloud access is secured by leveraging machine intelligence to uncover malicious domains, IPs, and URLs before they are even used in attacks. Once a malicious agent is discovered on one network, it is blacklisted across all customer networks. Machine learning is also used to detect anomalies in IT environments in order to safeguard the use of SaaS applications by adaptively learning user behaviour. Infrastructure-as-a-Service instances as well are safeguarded by using machine learning to discover advanced threats and malicious communications.

Intel – AI for cardiology treatment

Precision medicine for cancers requires the delivery of individually-adapted medical care based on the genetic characteristics of each patient. The last decade witnessed the development of high-throughput technologies such as next-generation sequencing, which paved their way in the field of oncology. While the cost of these technologies decreases, we are facing an exponential increase in the amount of data produced. In order to open the access to more and more patients to precision medicine-based therapies, healthcare providers have to rationalise both their data production and utilisation and this requires the implementation of the cuttingedge technology of high-performance computing and artificial intelligence.

Before taking a therapeutic decision based on the genome interpretation of a cancer, the physician can be presented with an overwhelming number of genes variants. In order to identify key actionable variants that can be targeted by treatments, the physician needs tools to sift through this large volume of variants. While the use of AI in genome interpretation is still nascent, it is growing rapidly as acting filter to dramatically reduce the number of variants, providing invaluable help to the physician. The mastering of high-performance computing methods on modern hardware infrastructure is becoming a key factor of the cancer genome interpretation process while being efficient, cost-effective and adjustable over time.

The pioneer collaboration initiated between the Curie Institute Bioinformatics platform and Intel aims at answering those challenges by defining a leading model in France and Europe. This collaboration will grant Institute Curie access to Intel experts for defining highperformance computing and artificial intelligence infrastructure and ensuring its optimisation in order to implement the Intel Genomics ecosystem partner solutions and best practices, for example the Broad Institute for Cancer Genomics pipeline optimisation. Also anticipated is the development of additional tailored tools needed to integrate and analyse heterogeneous biomedical data.

MSD – AI for healthcare professionals

MSD has launched, as part of its MSD Salute programme in Italy, a chatbot for physicians, powered by AI and machine learning. It has already achieved a large uptake with healthcare professionals in Italy. The programme’s sector of focus is immune-oncology.

From the MSD prospective, physicians are digital consumers looking for relevant information for their professional activity. Some key factors like the increase of media availability, mobile devices penetration and the decrease of time available, are resulting in a reduction of time spent navigating and searching on the web. Therefore users (and physicians with their pragmatic approach) read what they see and do not navigate as much but just ‘read and go’. This means that there is an urgent need to access content quickly, easily and efficiently.

The chatbot is developed in partnership with Facebook and runs on their Messenger app framework. As an easy and practical tool, it helps to establish a conversational relationship between the users. The MSD Italy ChatBot service is available only for registered physicians. Integration with Siri and other voice recognition systems is also worked on, to improve the human experience during the interaction with the chatbot. This initiative is a key item in MSD Italy’s digital strategy which focuses on new channels and touch-points with healthcare professionals, leveraging on new technologies.

Philips – AI in clinics and hospitals

With the clinical introduction of digital pathology, pioneered by Philips, it has become possible to implement more efficient pathology diagnostic workflows. This can help pathologists to streamline diagnostic processes, connect a team, even remotely, to enhance competencies and maximise use of resources, unify patient data for informed decision-making, and gain new insights by turning data into knowledge. Philips is working with PathAI to build deep learning applications. By analysing massive pathology data sets, we are developing algorithms aimed at supporting the detection of specific types of cancer and that inform treatment decisions.

Further, AI and machine learning for adaptive intelligence can also support quick action to address patient needs at the bedside. Manual patient health audits used to be timeconsuming, putting a strain on general ward staff. Nurses need to juggle a range of responsibilities: from quality of care to compliance with hospital standards. Information about the patient’s health was scattered across various records, making it even harder for nurses to focus their attention and take the right actions. Philips monitoring and notification systems assist nurses to detect a patient’s deterioration much quicker. All patient vital signs are automatically captured in one place to provide an Early Warning Score (EWS).

Microsoft – Machine learning for tumour detection and genome research

Microsoft’s Project InnerEye developed machine learning techniques for the automatic delineation of tumours as well as healthy anatomy in 3D radiological images. This technology helps to enable fast radiotherapy planning and precise surgery planning and navigation. Project InnerEye builds upon many years of research in computer vision and machine learning. The software learned how to mark organs and tumours up by training on a robust data set of images for patients that had been seen by experienced consultants.

The current process of marking organs and tumours on radiological images is done by medical practitioners and is very time consuming and expensive. Further, the process is a bottleneck to treatment – the tumour and healthy tissues must be delineated before treatment can begin. The InnerEye technology performs this task much more quickly than when done by hand by clinicians, reducing burdens on personnel and speeding up treatment.

The technology, however, does not replace the expertise of medical practitioners; it is designed to assist them and reduce the time needed for the task. The delineation provided by the technology is designed to be readily refined and adjusted by expert clinicians until completely satisfied with the results. Doctors maintain full control of the results at all times.

Further, Microsoft has partnered with St. Jude Children’s Research Hospital and DNANexus to develop a genomics platform that provides a database to enable researchers to identify how genomes differ. Researchers can inspect the data by disease, publication, gene mutation and also upload and test their own data using the bioinformatics tools. Researchers can progress their projects much faster and more cost-efficiently because the data and analysis run in the cloud, powered by rapid computing capabilities that do not require downloading anything.

Siemens – AI for Industry, Power Grids and Rail Systems

Siemens has been using smart boxes to bring older motors and transmissions into the digital age. These boxes contain sensors and communication interfaces for data transfer. By analysing the data, AI systems can draw conclusions regarding a machine’s condition and detect irregularities in order to make predictive maintenance possible.

AI is used also beyond industrial settings, for example to improve the reliability of power grids by making them smarter and providing the devices that control and monitor electrical networks with AI. This enables the devices to classify and localise disruptions in the grid. A special feature of this system is that the associated calculations are not performed centrally at a data centre, but de-centrally between the interlinked protection devices.

In cooperation with Deutsche Bahn, Siemens is running a pilot project for the predictive maintenance and repair of high-speed trains. Data analysts and software recognise patterns and trends from the vehicles’ operating data. Moreover, AI helps build optimised control centres for switch towers. From the billions of possible hardware configurations for a switch tower, the software selects options that fulfil all the requirements, including those regarding reliable operation.

Schneider Electric – AI for industry applications

Schneider Electric has used AI and machine learning in various sectors. In the oil and gas industry for example, machine learning is steering the operation of Realift rod pump control to monitor and configure pump settings and operations remotely, sending personnel onsite only when necessary for repair or maintenance – when Realift indicates that something has gone wrong. Anomalies in temperature and pressure, for instance, can flag potential problems, even issues brewing a mile below the surface. Intelligence edge devices can run analytics locally without having to tap the cloud — a huge deal for expensive, remote assets such as oil pumps.

To enable this solution an AI model is previously trained to recognise correct pump operation and also different types of failures a pump can experience, the AI model is deployed on a gateway at oil field for each pump and is fed with data collected at each pump stroke. Then, it outputs a prediction regarding the pump state. As we mimic the expert diagnostics, predictions can be easily validated, explained and interpreted.

Schneider Electric – Improving agriculture and farming with AI

Another example is in the agriculture sector, where Schneider Electric has proposed an AI solution for Waterforce, an irrigation solutions builder and water management company in New Zealand. Schneider Electric’ solution makes water use more efficient and effective in water use, saving up to 50% in energy costs, and provides remote monitoring capabilities that reduce the time farmers have to spend driving to inspect assets. The solution is able to collect data, from the weather forecast, pressure of pumps, temperatures, level of water, humidity of the ground, cleaning and selecting quality data, and preparing the data, in order to propose services such as fault diagnosis, performance benchmarking, recommendation and advise on operations.

AI and machine learning therefore represent a new way for humans and machines to work together – to learn about predictive tendencies and to solve complex problems. In the above examples, the challenges presented today in managing a process that requires tight control of temperatures, pressures, and liquid flows is quite complex and prone to error. Many variables need to be factored in to achieve a successful outcome – and the quality of the data that trains the AI algorithms could deliver very different results that the human brain should anyhow interpreted and guide. With the support of AI to make better operational decisions, critical factors such as safety, security, efficiency, productivity, and even profitability can be optimised in conjunction between machine/process and operator. This way, the training and combined skills from AI and expertise are a key success factor to deliver those values to Industry.

Canon – Application of automation in the office environment

Canon’s digital mailroom solution has been at the forefront of Robotic Process Automation (RPA) since it was first launched. A digital mailroom allows all incoming mail to be automatically captured, identified, validated and sent with relevant index data to the right systems or people. RPA technology is centred on removing the mundane to make lives easier. In the P2P world, RPA automates labour-intensive activities that require accessing multiple systems or that need to be audited for compliance.

Canon believes the next step in automation is the intelligent mailroom. The key challenge of the future will be the integration of digital and paper-based information into robust, effective and efficient processes. This means that organisations need more intelligent, digital mailroom solutions that enable data capture across every channel. One example of intelligent mailroom is the Multichannel Advanced Capture. This allows banks to enable customers to apply for an account minimising the amount of paper and using a mobile-friendly web page capturing the core details required. Automated checks on customers’ ID and credit history are made first. If all initial checks are valid, a second human check can be made. The bank is then presented with all the information required to make an informed decision on the application to open the bank account, based on applicable business rules as well as on (automatically) gathered historical business process knowledge.

SAS – Crowdsourcing and analysing data for endangered wildlife

The WildTrack Footprint Identification Technique (FIT) is a tool developed in partnership with SAS for non-invasive monitoring of endangered species through digital images of footprints. Measurements from these images are analysed by customised mathematical models that help to identify the species, individual, sex and age-class. AI could add the ability to adapt through progressive learning algorithms and tell an even more complete story.

Ordinary people would not necessarily be able to dart a rhino, but they can take an image of a footprint. WildTrack therefore has data coming in from everywhere. As this represents too much information to manage manually AI can automate repetitive learning through data, performing frequent, high-volume, computerised tasks reliably and without fatigue.

SAS – Using AI for real-time sports analytics

AI can also be used to analyse sports and football data. For example, SciSports models on-field movements using machine learning algorithms, which by nature improve on performing a task as they gain more experience. It works by automatically assigning a value to each action, such as a corner kick. Over time, these values change based on their success rate. A goal, for example, has a high value, but a contributing action – which may have previously had a low value – can become more valuable as the platform masters the game.

AI and machine learning will play an important role in the future of SciSports and football analytics in general. Existing mathematical models shape existing knowledge and insights in football, while AI and machine learning will make it possible to discover new connections that people would not make themselves.

Various other tools such as SAS Event Stream Processing and SAS Viya can then be utilised for real-time image recognition, with deep learning models, to distinguish between players, referees and the ball. The ability to deploy deep learning models in memory onto cameras and then do the inferencing in real time is cutting-edge science.

Google & TNO – AI for data analysis on traffic safety

TNO is one of the partners of InDeV, an international collaboration of researchers which was created to develop new ways of measuring traffic safety. Statistics about traffic safety were unreliable, insufficiently detailed, and hard to collect. Researchers often resort to filming busy intersections and manually reviewing the recording. This a time-intensive and expensive process. A single intersection needs to be monitored for three weeks with two cameras to create an estimation of its safety, adding up to six weeks of footage, which can take six weeks of work to analyse. Typically, less than one percent of the recorded material is actually of interest to researchers. The job of TNO is to apply machine learning to video of accident-prone hot spots to rate intersections on a scale according to their safety. With TNO’s neural network based on TensorFlow, researchers report that it takes only one hour to review footage that would previously have taken a week to inspect.

Advertisement

Research and Practice of AI Ethics: A Case Study Approach Juxtaposing Academic Discourse with Organisational Reality

• Original Research/Scholarship
• Open access
• Published: 08 March 2021
• Volume 27 , article number  16 , ( 2021 )

Cite this article

You have full access to this open access article

• Mark Ryan   ORCID: orcid.org/0000-0003-4850-0111 1 ,
• Josephina Antoniou 2 ,
• Laurence Brooks 3 ,
• Tilimbe Jiya 4 ,
• Kevin Macnish 5 &
• Bernd Stahl 3  

12k Accesses

18 Citations

7 Altmetric

Explore all metrics

This study investigates the ethical use of Big Data and Artificial Intelligence (AI) technologies (BD + AI)—using an empirical approach. The paper categorises the current literature and presents a multi-case study of 'on-the-ground' ethical issues that uses qualitative tools to analyse findings from ten targeted case-studies from a range of domains. The analysis coalesces identified singular ethical issues, (from the literature), into clusters to offer a comparison with the proposed classification in the literature. The results show that despite the variety of different social domains, fields, and applications of AI, there is overlap and correlation between the organisations’ ethical concerns. This more detailed understanding of ethics in AI + BD is required to ensure that the multitude of suggested ways of addressing them can be targeted and succeed in mitigating the pertinent ethical issues that are often discussed in the literature.

Similar content being viewed by others

The Ethical Implications of Artificial Intelligence (AI) For Meaningful Work

The rise of artificial intelligence – understanding the AI identity threat at the workplace

The potential of working hypotheses for deductive exploratory research

Avoid common mistakes on your manuscript.

Introduction

Big Data and Artificial Intelligence (BD + AI) are emerging technologies that offer great potential for business, healthcare, the public sector, and development agencies alike. The increasing impact of these two technologies and their combined potential in these sectors can be highlighted for diverse organisational aspects such as for customisation of organisational processes and for automated decision making. The combination of Big Data and AI, often in the form of machine learning applications, can better exploit the granularity of data and analyse it to offer better insights into behaviours, incidents, and risk, eventually aiming at positive organisational transformation.

Big Data offers fresh and interesting insights into structural patterns, anomalies, and decision-making in a broad range of different applications (Cuquet & Fensel, 2018 ), while AI provides predictive foresight, intelligent recommendations, and sophisticated modelling. The integration and combination of AI + BD offer phenomenal potential for correlating, predicting and prescribing recommendations in insurance, human resources (HR), agriculture, and energy, as well as many other sectors. While BD + AI provides a wide range of benefits, they also pose risks to users, including but not limited to privacy infringements, threats of unemployment, discrimination, security concerns, and increasing inequalities (O’Neil, 2016 ). Footnote 1 Adequate and timely policy needs to be implemented to prevent many of these risks occurring.

One of the main limitations preventing key decision-making for ethical BD + AI use is that there are few rigorous empirical studies carried out on the ethical implications of these technologies across multiple application domains. This renders it difficult for policymakers and developers to identify when ethical issues resulting from BD + AI use are only relevant for isolated domains and applications, or whether there are repeated/universal concerns which can be seen across different sectors. While the field lacks literature evaluating ethical issues Footnote 2 ‘on the ground’, there are even fewer multi-case evaluations.

This paper provides a cohesive multi-case study analysis across ten different application domains, including domains such as government, agriculture, insurance, and the media. It reviews ethical concerns found within these case studies to establish cross-cutting thematic issues arising from the implementation and use of BD + AI. The paper collects relevant literature and proposes a simple classification of ethical issues (short term, medium term, long term), which is then juxtaposed with the ethical concerns highlighted from the multiple-case study analysis. This multiple-case study analysis of BD + AI offers an understanding of current organisational practices.

The work described in this paper makes an important contribution to the literature, based on its empirical findings. By presenting the ethical issues across an array of application areas, the paper provides much-needed rigorous empirical insight into the social and organisational reality of ethics of AI + BD. Our empirical research brings together a collection of domains that gives a broad oversight about issues that underpin the implementation of AI. Through its empirical insights the paper provides a basis for a broader discussion of how these issues can and should be addressed.

This paper is structured in six main sections: this introduction is followed by a literature review, which allows for an integrated review of ethical issues, contrasting them with those found in the cases. This provides the basis for a categorisation or classification of ethical issues in BD + AI. The third section contains a description of the interpretivist qualitative case study methodology used in this paper. The subsequent section provides an overview of the organisations participating in the cases to contrast similarities and divisions, while also comparing the diversity of their use of BD + AI. Footnote 3 The fifth section provides a detailed analysis of the ethical issues derived from using BD + AI, as identified in the cases. The concluding section analyses the differences between theoretical and empirical work and spells out implications and further work.

Literature Review

An initial challenge that any researcher faces when investigating ethical issues of AI + BD is that, due to the popularity of the topic, there is a vast and rapidly growing literature to be considered. Ethical issues of AI + BD are covered by a number of academic venues, including some specific ones such as the AAAI/ACM Conference on AI, Ethics, and Society ( https://dl.acm.org/doi/proceedings/10.1145/3306618 ), policy initiative and many publicly and privately financed research reports (Whittlestone, Nyrup, Alexandrova, Dihal, & Cave, 2019 ). Initial attempts to provide overviews of the area have been published (Jobin, 2019 ; Mittelstadt, Allo, Taddeo, Wachter, & Floridi, 2016 ), but there is no settled view on what counts as an ethical issue and why. In this paper we aim to provide a broad overview of issues found through the case studies. This paper puts forward what are commonly perceived to be ethical issues within the literature or concerns that have ethical impacts and repercussions. We explicitly do not apply a particular philosophical framework of ethics but accept as ethical issues those issues that we encounter in the literature. This review is based on an understanding of the current state of the literature by the paper's authors. It is not a structured review and does not claim comprehensive coverage but does share some interesting insights.

To be able to undertake the analysis of ethical issues in our case studies, we sought to categorise the ethical issues found in the literature. There are potentially numerous ways of doing so and our suggestion does not claim to be authoritative. Our suggestion is to order ethical issues in terms of their temporal horizon, i.e., the amount of time it is likely to take to be able to address them. Time is a continuous variable, but we suggest that it is possible to sort the issues into three clusters: short term, medium term, and long term (see Fig.  1 ).

Temporal horizon for addressing ethical issues

As suggested by Baum ( 2017 ), it is best to acknowledge that there will be ethical issues and related mitigating activities that cannot exclusively fit in as short, medium or long term.

ather than seeing it as an authoritative classification, we see this as a heuristic that reflects aspects of the current discussion. One reason why this categorisation is useful is that the temporal horizon of ethical issues is a potentially useful variable, with companies often being accused of favouring short-term gains over long-term benefits. Similarly, short-term issues must be able to be addressed on the local level for short-term fixes to work.

Short-term issues

These are issues for which there is a reasonable assumption that they are capable of being addressed in the short term. We do not wish to quantify what exactly counts as short term, as any definition put forward will be contentious when analysing the boundaries and transition periods. A better definition of short term might therefore be that such issues can be expected to be successfully addressed in technical systems that are currently in operation or development. Many of the issues we discuss under the heading of short-term issues are directly linked to some of the key technologies driving the current AI debate, notably machine learning and some of its enabling techniques and approaches such as neural networks and reinforcement learning.

Many of the advantages promised by BD + AI involve the use of personal data, data which can be used to identify individuals. This includes health data; customer data; ANPR data (Automated Number Plate Recognition); bank data; and even includes data about farmers’ land, livestock, and harvests. Issues surrounding privacy and control of data are widely discussed and recognized as major ethical concerns that need to be addressed (Boyd & Crawford, 2012 ; Tene & Polonetsky, 2012 , 2013 ; Mittelstadt, Allo, Taddeo, Wachter, & Floridi, 2016 ; Jain, Gyanchandani, & Khare, 2016 ; Mai, 2016 ; Macnish, 2018 ). The concern surrounding privacy can be put down to a combination of a general level of awareness of privacy issues and the recently-introduced General Data Protection Regulation (GDPR). Closely aligned with privacy issues are those relating to transparency of processes dealing with data, which can often be classified as internal, external, and deliberate opaqueness (Burrell, 2016 ; Lepri, Staiano, Sangokoya, Letouzé, & Oliver, 2017 ; Mittelstadt, Allo, Taddeo, Wachter, & Floridi, 2016 ).

The Guidelines for Trustworthy AI Footnote 4 were released in 2018 by the High-Level Expert Group on Artificial Intelligence (AI HLEG Footnote 5 ), and address the need for technical robustness and safety, including accuracy, reproducibility, and reliability. Reliability is further linked to the requirements of diversity, fairness, and social impact because it addresses freedom from bias from a technical point of view. The concept of reliability, when it comes to BD + AI, refers to the capability to verify the stability or consistency of a set of results (Bush, 2012 ; Ferraggine, Doorn, & Rivera, 2009 ; Meeker and Hong, 2014 ).

If a technology is unreliable, error-prone, and unfit-for-purpose, adverse ethical issues may result from decisions made by the technology. The accuracy of recommendations made by BD + AI is a direct consequence of the degree of reliability of the technology (Barolli, Takizawa, Xhafa, & Enokido, 2019 ). Bias and discrimination in algorithms may be introduced consciously or unconsciously by those employing the BD + AI or because of algorithms reflecting pre-existing biases (Baroccas and Selbst, 2016 ). Examples of bias have been documented often reflecting “an imbalance in socio-economic or other ‘class’ categories—ie, a certain group or groups are not sampled as much as others or at all” (Panch et al., 2019 ). have the potential to affect levels of inequality and discrimination, and if biases are not corrected these systems can reproduce existing patterns of discrimination and inherit the prejudices of prior decision makers (Barocas & Selbst, 2016 , p. 674). An example of inherited prejudices is documented in the United States, where African-American citizens, more often than not, have been given longer prison sentences than Caucasians for the same crime.

Medium-term issues

Medium-term issues are not clearly linked to a particular technology but typically arise from the integration of AI techniques including machine learning into larger socio-technical systems and contexts. They are thus related to the way life in modern societies is affected by new technologies. These can be based on the specific issues listed above but have their main impact on the societal level. The use of BD + AI may allow individuals’ behaviour to be put under scrutiny and surveillance , leading to infringements on privacy, freedom, autonomy, and self-determination (Wolf, 2015 ). There is also the possibility that the increased use of algorithmic methods for societal decision-making may create a type of technocratic governance (Couldry & Powell, 2014 ; Janssen & Kuk, 2016 ), which could infringe on people’s decision-making processes (Kuriakose & Iyer, 2018 ). For example, because of the high levels of public data retrieval, BD + AI may harm people’s freedom of expression, association, and movement, through fear of surveillance and chilling effects (Latonero, 2018 ).

Corporations have a responsibility to the end-user to ensure compliance, accountability, and transparency of their BD + AI (Mittelstadt, Allo, Taddeo, Wachter, & Floridi, 2016 ). However, when the source of a problem is difficult to trace, owing to issues of opacity, it becomes challenging to identify who is responsible for the decisions made by the BD + AI. It is worth noting that a large-scale survey in Australia in 2020 indicated that 57.9% of end-users are not at all confident that most companies take adequate steps to protect user data. The significance of understanding and employing responsibility is an issue targeted in many studies (Chatfield et al., 2017 ; Fothergill et al., 2019 ; Jirotka et al., 2017 ; Pellé & Reber, 2015 ). Trust and control over BD + AI as an issue is reiterated by a recent ICO report demonstrating that most UK citizens do not trust organisations with their data (ICO, 2017 ).

Justice is a central concern in BD + AI (Johnson, 2014 , 2018 ). As a starting point, justice consists in giving each person his or her due or treating people equitably (De George, p. 101). A key concern is that benefits will be reaped by powerful individuals and organisations, while the burden falls predominantly on poorer members of society (Taylor, 2017 ). BD + AI can also reflect human intentionality, deploying patterns of power and authority (Portmess & Tower, 2015 , p. 1). The knowledge offered by BD + AI is often in the hands of a few powerful corporations (Wheeler, 2016 ). Power imbalances are heightened because companies and governments can deploy BD + AI for surveillance, privacy invasions and manipulation, through personalised marketing efforts and social control strategies (Lepri, Staiano, Sangokoya, Letouzé, & Oliver, 2017 , p. 11). They play a role in the ascent of datafication, especially when specific groups (such as corporate, academic, and state institutions) have greater unrestrained access to big datasets (van Dijck, 2014 , p. 203).

Discrimination , in BD + AI use, can occur when individuals are profiled based on their online choices and behaviour, but also their gender, ethnicity and belonging to specific groups (Calders, Kamiran, & Pechenizkiy, 2009 ; Cohen et al., 2014 ; and Danna & Gandy, 2002 ). Data-driven algorithmic decision-making may lead to discrimination that is then adopted by decision-makers and those in power (Lepri, Staiano, Sangokoya, Letouzé, & Oliver, 2017 , p. 4). Biases and discrimination can contribute to inequality . Some groups that are already disadvantaged may face worse inequalities, especially if those belonging to historically marginalised groups have less access and representation (Barocas & Selbst, 2016 , p. 685; Schradie, 2017 ). Inequality-enhancing biases can be reproduced in BD + AI, such as the use of predictive policing to target neighbourhoods of largely ethnic minorities or historically marginalised groups (O’Neil, 2016 ).

BD + AI offers great potential for increasing profit, reducing physical burdens on staff, and employing innovative sustainability practices (Badri, Boudreau-Trudel, & Souissi, 2018 ). They offer the potential to bring about improvements in innovation, science, and knowledge; allowing organisations to progress, expand, and economically benefit from their development and application (Crawford et al., 2014 ). BD + AI are being heralded as monumental for the economic growth and development of a wide diversity of industries around the world (Einav & Levin, 2014 ). The economic benefits accrued from BD + AI may be the strongest driver for their use, but BD + AI also holds the potential to cause economic harm to citizens and businesses or create other adverse ethical issues (Newman, 2013 ).

However, some in the literature view the co-development of employment and automation as somewhat naïve outlook (Zuboff, 2015 ). BD + AI companies may benefit from a ‘post-labour’ automation economy, which may have a negative impact on the labour market (Bossman, 2016 ), replacing up to 47% of all US jobs within the next 20 years (Frey & Osborne, 2017 ). The professions most at risk of affecting employment correlated with three of our case studies: farming, administration support and the insurance sector (Frey & Osborne, 2017 ).

Long-term issues

Long-term issues are those pertaining to fundamental aspects of nature of reality, society, or humanity. For example, that AI will develop capabilities far exceeding human beings (Kurzweil, 2006 ). At this point, sometimes called the ‘ singularity ’ machines achieve human intelligence, are expected to be able to improve on themselves and thereby surpass human intelligence and become superintelligent (Bostrom, 2016 ). If this were to happen, then it might have dystopian consequences for humanity as often depicted in science fiction. Also, it stands to reason that the superintelligent, or even just the normally intelligent machines may acquire a moral status.

It should be clear that these expectations are not universally shared. They refer to what is often called ‘ artificial general intelligence’ (AGI), a set of technologies that emulate human reasoning capacities more broadly. Footnote 6

Furthermore, if we may acquire new capabilities, e.g. by using technical implants to enhance human nature. The resulting being might be called a transhuman , the next step of human evolution or development. Again, it is important to underline that this is a contested idea (Livingstone, 2015 ) but one that has increasing traction in public discourse and popular science accounts (Harari, 2017 ).

We chose this distinction of three groups of issues for understanding how mitigation strategies within organisations can be contextualised. We concede that this is one reading of the literature and that many others are possible. In this account of the literature we tried to make sense of the current discourse to allow us to understand our empirical findings which are introduced in the following sections.

Case Study Methodology

Despite the impressive amount of research undertaken on ethical issues of AI + BD (e.g. Mittelstadt, Allo, Taddeo, Wachter, & Floridi, 2016 ; Zwitter, 2014 ), there are few case studies exploring such issues. This paper builds upon this research and employs an interpretivist methodology to do so, focusing on how, what, and why questions relevant to the ethical use of BD + AI (Walsham, 1995a , b ). The primary research questions for the case studies were: How do organisations perceive ethical concerns related to BD + AI and in what ways do they deal with them?

We sought to elicit insights from interviews, rather than attempting to reach an objective truth about the ethical impacts of BD + AI. The interpretivist case study approach (Stake 2003) allowed the researchers ‘to understand ‘reality’ as the blending of the various (and sometimes conflicting) perspectives which coexist in social contexts, the common threads that connect the different perspectives and the value systems that give rise to the seeming contradictions and disagreements around the topics discussed. Whether one sees this reality as static (social constructivism) or dynamic (social constructionism) was also a point of consideration, as they both belong in the same “family” approach where methodological flexibility is as important a value as rigour’ (XXX).

Through extensive brainstorming within the research team, and evaluations of relevant literature, 16 social application domains were established as topics for case study analysis. Footnote 7 The project focused on ten out of these application domains in accordance with the partners’ competencies. The case studies have covered ten domains, and each had their own unique focus, specifications, and niches, which added to the richness of the evaluations (Table 1 ).

The qualitative analysis approach adopted in this study focused on these ten standalone operational case studies that were directly related to the application domains presented in Table 1 . These individual case studies provide valuable insights (Yin, 2014 , 2015 ); however, a multiple-case study approach offers a more comprehensive analysis of ethical issues related to BD + AI use (Herriott & Firestone, 1983 ). Thus, this paper adopts a multiple-case study methodology to identify what insights can be obtained from the ten cases, identifies whether any generalisable understandings can be retrieved, and evaluates how different organisations deal with issues pertaining to BD + AI development and use. The paper does not attempt to derive universal findings from this analysis, in line with the principles of interpretive research, but further attempts to gain an in-depth understanding of the implications of selected BD + AI applications.

The data collection was guided by specific research questions identified through each case, including five desk research questions (see appendix 1); 24 interview questions (see appendix 2); and a checklist of 17 potential ethical issues, developed by the project leader Footnote 8 (see appendix 3). A thematic analysis framework was used to ‘highlight, expose, explore, and record patterns within the collected data. The themes were patterns across data sets that were important to describe several ethical issues which arise through the use of BD  +  AI across different types of organisations and application domains’ (XXX).

A workshop was then held after the interviews were carried out. The workshop brought together the experts in the case study team to discuss their findings. This culminated in 26 ethical issues Footnote 9 that were inductively derived from the data collected throughout the interviews (see Fig.  2 and Table 3). Footnote 10 In order to ensure consistency and rigour in the multiple-case study approach, researchers followed a standardised case study protocol (Yin, 2014 ). Footnote 11

The Prevalence of Ethical Issues in the Case Studies

Thirteen different organisations were interviewed for 10 case studies, consisting of 22 interviews in total. Footnote 12 These ranged from 30 min to 1 ½ hours in-person or Skype interviews. The participants that were selected for interviews represented a very broad range of application domains and organisations that use BD + AI. The case study organisations were selected according to their relevance to the overall case study domains and considering their fit with the domains and likelihood of providing interesting insights. The interviewees were then selected according to their ability to explain their BD + AI and its role in their organisation. In addition to interviews, a document review provided supporting information about the organisation. Thus, websites and published material were used to provide background to the research.

Findings: Ten Case Studies

This section gives a brief overview of the cases, before analysing their similarities and differences. It also highlights the different types of BD + AI being used, and the types of data used by the BD + AI in the case study organisations, before conducting an ethical analysis of the cases. Table 2 presents an overview of the 10 cases to show the roles of the interviewees, the focus of the technologies being used, and the data retrieved by each organisation’s BD + AI. All interviews were conducted in English.

The types of organisations that were used in the case studies varied extensively. They included start-ups (CS10), niche software companies (CS1), national health insurers (Organisation X in CS6), national energy providers (CS7), chemical/agricultural multinational (CS3), and national (CS9) and international (CS8) telecommunications providers. The case studies also included public (CS2, Organisation 1 and 4 in CS4) and semi-public (Organisation 2 in CS4) organisations, as well as a large scientific research project (CS5).

The types of individuals interviewed also varied extensively. For example, CS6 and CS7 did not have anyone with a specific technical background, which limited the possibility of analysing issues related to the technology itself. Some case studies only had technology experts (such as CS1, CS8, and CS9), who mostly concentrated on technical issues, with much less of a focus on ethical concerns. Other case studies had a combination of both technical and policy-focused experts (i.e. CS3, CS4, and CS5). Footnote 13

Therefore, it must be made fundamentally clear that we are not proposing that all of the interviewees were authorities in the field, or that even collectively they represent a unified authority on the matter, but instead, that we are hoping to show what are the insights and perceived ethical issues of those currently working with AI on the ground view as ethical concerns. While the paper is presenting the ethical concerns found within an array of domains, we do not claim that any individual case study is representative of their entire industry, but instead, our intent was to capture a wide diversity of viewpoints, domains, and applications of AI, to encompass a broad amalgamation of concerns. We should also state that this is not a shortcoming of the study but that it is the normal approach that social science often takes.

The diversity of organisations and their application focus areas also varied. Some organisations focused more so on the Big Data component of their AI, while others more strictly on the AI programming and analytics. Even when organisations concentrated on a specific type of BD + AI, such as Big Data, its use varied immensely, including retrieval (CS1), analysis (CS2), predictive analytics (CS10), and transactional value (Organisation 2 in CS4). Some domains adopted BD + AI earlier and more emphatically than others (such as communications, healthcare, and insurance). Also, the size, investment, and type of organisation played a part in the level of BD + AI innovation (for example, the two large multinationals in CS3 and CS8 had well-developed BD + AI).

The maturity level of BD + AI was also determined by how it was integrated, and its importance, within an organisation. For instance, in organisations where BD + AI were fundamental for the success of the business (e.g. CS1 and CS10), they played a much more important role than in companies where there was less of a reliance (e.g. CS7). In some organisations, even when BD + AI was not central to success, the level of development was still quite advanced because of economic investment capabilities (e.g. CS3 and CS8).

These differences provided important questions to ask throughout this multi-case study analysis, such as: Do certain organisations respond to ethical issues relating to BD + AI in a certain way? Does the type of interviewee affect the ethical issues discussed—e.g. case studies without technical experts, those that only had technical experts, and those that had both? Does the type of BD + AI used impact the types of ethical issues discussed? What significance does the type of data retrieved have on ethical issues identified by the organisations? These inductive ethical questions provided a template for the qualitative analysis in the following section.

Ethical Issues in the Case Studies

Based on the interview data, the ethical issues identified in the case studies were grouped into six specific thematic sections to provide a more conducive, concise, and pragmatic methodology. Those six sections are: control of data, reliability of data, justice, economic issues, role of organisations, and individual freedoms. From the 26 ethical issues, privacy was the only ethical issue addressed in all 10 case studies, which was not surprising because it has received a great deal of attention recently because of the GDPR. Also, security, transparency, and algorithmic bias are regularly discussed in the literature, so we expected them to be significant issues across many of the cases. However, there were many issues that received less attention in the literature—such as access to BD + AI, trust, and power asymmetries—which were discussed frequently in the interviews. In contrast to this, there were ethical issues that were heavily discussed in the literature which received far less attention in the interviews, such as employment, autonomy, and criminal or malicious use of BD + AI (Fig.  2 ).

The ethical analysis was conducted using a combination of literature reviews and interviews carried out with stakeholders. The purpose of the interviews was to ensure that there were no obvious ethical issues faced by stakeholders in their day-to-day activities which had been missed in the academic literature. As such, the starting point was not an overarching normative theory, which might have meant that we looked for issues which fit well with the theory but ignored anything that fell outside of that theory. Instead the combined approach led to the identification of the 26 ethical issues, each labelled based on particular words or phrases used in the literature or by the interviewees. For example, the term "privacy" was used frequently and so became the label for references to and instances of privacy-relevant concerns. In this section we have clustered issues together based on similar problems faced (e.g. accuracy of data and accuracy of algorithms within the category of ‘reliability of data’).

In an attempt to highlight similar ethical issues and improve the overall analysis to better capture similar perspectives, the research team decided to use the method of clustering, a technique often used in data mining to efficiently group similar elements together. Through discussion in the research team, and bearing in mind that the purpose of the clustering process was to form clusters that would enhance understanding of the impact of these ethical issues, we arrived at the following six clusters: the control of data (covering privacy, security, and informed consent); the reliability of data (accuracy of data and accuracy of algorithms); justice (power asymmetries, justice, discrimination, and bias); economic issues (economic concerns, sustainability, and employment); the role of organisations (trust and responsibility); and human freedoms (autonomy, freedom, and human rights). Both the titles and the precise composition of each cluster of issues are the outcome of a reasoned agreement of the research team. However, it should be clear that we could have used different titles and different clustering. The point is not that each cluster forms a distinct group of ethical issues, independent from any other. Rather the ethical issues faced overlap and play into one another, but to present them in a manageable format we have opted to use this bottom-up clustering approach.

Human Freedoms

An interviewee from CS10 stated that they were concerned about human rights because they were an integral part of the company’s ethics framework. This was beneficial to their business because they were required to incorporate human rights to receive public funding by the Austrian government. The company ensured that they would not grant ‘full exclusivity on generated social unrest event data to any single party, unless the data is used to minimise the risk of suppression of unrest events, or to protect the violation of human rights’ (XXX). The company demonstrates that while BD + AI has been criticised for infringing upon human rights in the literature, they also offer the opportunity to identify and prevent human rights abuses. The company’s moral framework definitively stemmed from regulatory and funding requirements, which lends itself to the benefit of effective ethical top-down approaches, which is a divisive topic in the literature, with diverging views about whether top-down or bottom-up approaches are better options for improved AI ethics.

Trust & Responsibility

Responsibility was a concern in 5 of the case studies, confirming the importance it is given in the literature (see Sect.  3 ). Trust appeared in seven of the case studies. The cases focused on concerns found in the literature, such as BD + AI use in policy development, public distrust about automated decision-making and the integrity of corporations utilising datafication methods (van Dijck 2014 ).

Trust and control over BD + AI were an issue throughout the case studies. The organisation from the predictive intelligence case study (CS10) identified that their use of social media data raised trust issues. They converged with perspectives found in the literature that when people feel disempowered to use or be part of the BD + AI development process, they tend to lose trust in the BD + AI (Accenture, 2016 , 2017 ). In CS6, stakeholders (health insurers) trusted the decisions made by BD + AI when they were engaged and empowered to give feedback on how their data was used. Trust is enhanced when users can refuse the use of their data (CS7), which correlates with the literature. Companies discussed the benefits of establishing trustworthy relationships. For example, in CS9, they have “ been trying really hard to avoid the existence of fake [mobile phone] base stations, because [these raise] an issue with the trust that people put in their networks” (XXX).

Corporations need to determine the objective of the data analysis (CS3), what data is required for the BD + AI to work (CS2), and accountability for when it does not work as intended or causes undesirable outcomes (CS4). The issue here is whether the organisation takes direct responsibility for these outcomes, or, if informed consent has been given, can responsibility be shared with the granter of consent (CS3). The cases also raised the question of ‘responsible to whom’, the person whose data is being used or the proxy organisation who has provided data (CS6). For example, in the insurance case study, the company stated that they only had a responsibility towards the proxy organisation and not the sources of the data. All these issues are covered extensively in the literature in most application domains.

Control of Data

Concerns surrounding the control of data for privacy reasons can be put down to a general awareness of privacy issues in the press, reinforced by the recently-introduced GDPR. This was supported in the cases, where interviewees expressed the opinion that the GDPR had raised general awareness of privacy issues (CS1, CS9) or that it had lent weight to arguments concerning the importance of privacy (CS8).

The discussion of privacy ranged from stressing that it was not an issue for some interviewees, because there was no personal information in the data they used (CS4), to its being an issue for others, but one which was being dealt with (CS2 and CS8). One interviewee (CS5) expressed apprehension that privacy concerns conflicted with scientific innovation, introducing hitherto unforeseen costs. This view is not uncommon in scientific and medical innovation, where harms arising from the use of anonymised medical data are often seen as minimal and the potential benefits significant (Manson & O’Neill, 2007 ). In other cases (CS1), there was a confusion between anonymisation (data which cannot be traced back to the originating source) and pseudonymisation (where data can be traced back, albeit with difficulty) of users’ data. A common response from the cases was that providing informed consent for the use of personal data waived some of the rights to privacy of the user.

Consent may come in the form of a company contract Footnote 14 or an individual agreement. Footnote 15 In the former, the company often has the advantage of legal support prior to entering a contract and so should be fully aware of the information provided. In individual agreements, though, the individual is less likely to be legally supported, and so may be at risk of exploitation through not reading the information sufficiently (CS3), or of responding without adequate understanding (CS9). In one case (CS5), referring to anonymised data, consent was implied rather than given: the interviewee suggested that those involved in the project may have contributed data without giving clear informed consent. The interviewee also noted that some data may have been shared without the permission, or indeed knowledge, of those contributing individuals. This was acknowledged by the interviewee as a potential issue.

In one case (CS6), data was used without informed consent for fraud detection purposes. The interviewees noted that their organisation was working within the parameters of national and EU legislation, which allows for non-consensual use of data for these ends. One interviewee in this case stated that informed consent was sought for every novel use of the data they held. However, this was sought from the perceived owner of the data (an insurance company) rather than from the originating individuals. This case demonstrates how people may expect their data to be used without having a full understanding of the legal framework under which the data are collected. For example, data relating to individuals may legally be accessed for fraud detection without notifying the individual and without relying on the individual’s consent.

This use of personal data for fraud detection in CS6 also led to concerns regarding opacity. In both CS6 and CS10 there was transparency within the organisations (a shared understanding among staff as to the various uses of the data) but that did not extend to the public outside those organisations. In some cases (CS5) the internal transparency/external opacity meant that those responsible for developing BD + AI were often hard to meet. Of those who were interviewed in CS5, many did not know the providence of the data or the algorithms they were using. Equally, some organisations saw external opacity as integral to the business environment in which they were operating (CS9, CS10) for reasons of commercial advantage. The interviewee in CS9 cautioned that this approach, coupled with a lack of public education and the speed of transformation within the industry, would challenge any meaningful level of public accountability. This would render processes effectively opaque to the public, despite their being transparent to experts.

Reliability of Data

There can be multiple sources of unreliability in BD + AI. Unreliability originating from faults in the technology can lead to algorithmic bias, which can cause ethical issues such as unfairness, discrimination, and general negative social impact (CS3 and CS6). Considering algorithmic bias as a key input to data reliability, there exist two types of issues that may need to be addressed. Primarily, bias may stem from the input data, referred to as training data, if such data excludes adequate representation of the world, e.g. gender-biased datasets (CS6). Secondly, an inadequate representation of the world may be the result of lack of data, e.g. a correctly designed algorithm to learn from and predict a rare disease, may not have sufficient representative data to achieve correct predictions (CS5). In either case the input data are biased and may result in inaccurate decision-making and recommendations.

The issues of reliability of data stemming from data accuracy and/or algorithmic bias, may escalate depending on their use, as for example in predictive or risk-assessment algorithms (CS10). Consider the risks of unreliable data in employee monitoring situations (CS1), detecting pests and diseases in agriculture (CS3), in human brain research (CS5) or cybersecurity applications (CS8). Such issues are not singular in nature but closely linked to other ethical issues such as information asymmetries, trust, and discrimination. Consequently, the umbrella issue of reliability of data must be approached from different perspectives to ensure the validity of the decision-making processes of the BD + AI.

Data may over-represent some people or social groups who are likely to be already privileged or under-represent disadvantaged and vulnerable groups (CS3). Furthermore, people who are better positioned to gain access to data and have the expertise to interpret them may have an unfair advantage over people devoid of such competencies. In addition, BD + AI can work as a tool of disciplinary power, used to evaluate people’s conformity to norms representing the standards of disciplinary systems (CS5). We focus on the following aspects of justice in our case study analysis: power asymmetries, discrimination, inequality, and access.

The fact that issues of power can arise in public as well as private organisations was discussed in our case studies. The smart city case (CS4) showed that the public organisations were aware of potential problems arising from companies using public data and were trying to put legal safeguards in place to avoid such misuse. As a result of misuse, there is the potential that cities, or the companies with which they contract, may use data in harmful or discriminatory ways. Our case study on the use of BD + AI in scientific research showed that the interviewees were acutely aware of the potential of discrimination (CS10). They stated that biases in the data may not be easy to identify, and may lead to misclassification or misinterpretation of findings, which may in turn skew results. Discrimination refers to the recognition of difference, but it may also refer to unjust treatment of different categories of people based on their gender, sex, religion, race, class, or disability. BD + AI are often employed to distinguish between different cases, e.g. between normal and abnormal behaviour in cybersecurity. Determining whether such classification entails discrimination in the latter sense can be difficult, due to the nature of the data and algorithms involved.

Examples of potential inequality based on BD + AI could be seen in several case studies. The agricultural case (CS3) highlighted the power differential between farmers and companies with potential implications for inequality, but also the global inequality between farmers, linked to farming practices in different countries (CS3). Subsistence farmers in developing countries, for example, might find it more difficult to benefit from these technologies than large agro-businesses. The diverging levels of access to BD + AI entail different levels of ability to benefit from them and counteract possible disadvantages (CS3). Some companies restrict access to their data entirely, and others sell access at a fee, while others offer small datasets to university-based researchers (Boyd & Crawford, 2012 , p. 674).

Economic Issues

One economic impact of BD + AI outlined in the agriculture case study (CS3) focused on whether this technology, and their ethical implementation, were economically affordable. If BD + AI could not improve economic efficiency, they would be rejected by the end-user, whether they were more productive, sustainable, and ethical options. This is striking, as it raises a serious challenge for the AI ethics literature and industry. It establishes that no matter how well intentioned and principled AI ethics guidelines and charters are, unless their implementation can be done in an economically viable way, their implementation will be challenged and resisted by those footing the bill.

The telecommunications case study (CS9) focused on how GDPR legislation may economically impact businesses using BD + AI by creating disparities in competitiveness between EU and non-EU companies developing BD + AI. Owing to the larger data pools of the latter, their BD + AI may prove to be more effective than European-manufactured alternatives, which cannot bypass the ethical boundaries of European law in the same way (CS8). This is something that is also being addressed in the literature and is a very serious concern for the future profitability and development of AI in Europe (Wallace & Castro, 2018 ). The literature notes additional issues in this area that were not covered in the cases. There is the potential that the GDPR will increase costs of European AI companies by having to manually review algorithmic decision-making; the right to explanation could reduce AI accuracy; and the right to erasure could damage AI systems (Wallace & Castro, 2018 , p. 2).

One interviewee stated that public–private BD + AI projects should be conducted in a collaborative manner, rather than a sale-of-service (CS4). However, this harmonious partnership is often not possible. Another interviewee discussed the tension between public and private interests on their project—while the municipality tried to focus on citizen value, the ICT company focused on the project’s economic success. The interviewee stated that the project would have terminated earlier if it were the company’s decision, because it was unprofitable (CS4). This is a huge concern in the literature, whereby private interests will cloud, influence, and damage public decision-making within the city because of their sometimes-incompatible goals (citizen value vs. economic growth) (Sadowski & Pasquale, 2015 ). One interviewee said that the municipality officials were aware of the problems of corporate influence and thus are attempting to implement the approach of ‘data sovereignty’ (CS2).

During our interviews, some viewed BD + AI as complementary to human employment (CS3), collaborative with such employment (CS4), or as a replacement to employment (CS6). The interviewees from the agriculture case study (CS3) stated that their BD + AI were not sufficiently advanced to replace humans and were meant to complement the agronomist, rather than replace them. However, they did not indicate what would happen when the technology is advanced enough, and it becomes profitable to replace the agronomist. The insurance company interviewee (CS6) stated that they use BD + AI to reduce flaws in personal judgment. The literature also supports this viewpoint, where BD + AI is seen to offer the potential to evaluate cases impartially, which is beneficial to the insurance industry (Belliveau, Gray, & Wilson, 2019 ). Footnote 16 The interviewee reiterated this and also stated that BD + AI would reduce the number of people required to work on fraud cases. The interviewee stated that BD + AI are designed to replace these individuals, but did not indicate whether their jobs were secure or whether they would be retrained for different positions, highlighting a concern found in the literature about the replacement and unemployment of workers by AI (Bossman, 2016 ). In contrast to this, a municipality interviewee from CS4 stated that their chat-bots are used in a collaborative way to assist customer service agents, allowing them to concentrate on higher-level tasks, and that there are clear policies set in place to protect their jobs.

Sustainability was only explicitly discussed in two interviews (CS3 and CS4). The agriculture interviewees stated that they wanted to be the ‘first’ to incorporate sustainability metrics into agricultural BD + AI, indicating a competitive and innovative rationale for their company (CS3). Whereas the interviewee from the sustainable development case study (CS4) stated that their goal of using BD + AI was to reduce Co2 emissions and improve energy and air quality. He stated that there are often tensions between ecological and economic goals and that this tension tends to slow down the efforts of BD + AI public–private projects—an observation also supported by the literature (Keeso, 2014 ). This tension between public and private interests in BD + AI projects was a recurring issue throughout the cases, which will be the focus of the next section on the role of organisations.

Discussion and Conclusion

The motivation behind this paper is to come to a better understanding of ethical issues related to BD + AI based on a rich empirical basis across different application domains. The exploratory and interpretive approach chosen for this study means that we cannot generalise from our research to all possible examples of BD + AI, but it does allow us to generalise to theory and rich insights (Walsham, 1995a , b , 2006 ). These theoretical insights can then provide the basis for further empirical research, possibly using other methods to allow an even wider set of inputs to move beyond some of the limitations of the current study.

Organisational Practice and the Literature

The first point worth stating is that there is a high level of consistency both among the case studies and between cases and literature. Many of the ethical issues identified cut across the cases and are interpreted in similar ways by different stakeholders. The frequency distribution of ethical issues indicates that very few, if any, issues are relevant to all cases but many, such as privacy, have a high level of prevalence. Despite appearing in all case studies, privacy was not seen as overly problematic and could be dealt with in the context of current regulatory principles (GDPR). Most of the issues that we found in the literature (see Sect.  2 ) were also present in the case studies. In addition to privacy and data protection, this included accuracy, reliability, economic and power imbalances, justice, employment, discrimination and bias, autonomy and human rights and freedoms.

Beyond the general confirmation of the relevance of topics discussed in the literature, though, the case studies provide some further interesting insights. From the perspective of an individual case some societal factors are taken for granted and outside of the control of individual actors. For example, intellectual property regimes have significant and well-recognised consequences for justice, as demonstrated in the literature. However, there is often little that individuals or organisations can do about them. Even in cases where individuals may be able to make a difference and the problem is clear, it is not always obvious how to do this. Some well-publicised discrimination cases may be easy to recognise, for example where an HR system discriminates against women or where a facial recognition system discriminates against black people. But in many cases, it may be exceedingly difficult to recognise discrimination where it is not clear how a person is discriminated against. If, for example, an image-based medical diagnostic system leads to disadvantages for people with genetic profiles, this may not be easy to identify.

With regards to the classification of the literature suggested in Sect.  2 along the temporal dimension, we can see that the attention of the case study respondents seems to be correlated to the temporal horizon of the issues. The issues we see as short-term figures most prominently, whereas the medium-term issues, while still relevant and recognisable, appear to be less pronounced. The long-term questions are least visible in the cases. This is not very surprising, as the short-term issues are those that are at least potentially capable of being addressed relatively quickly and thus must be accessible on the local level. Organisations deploying or using AI therefore are likely to have a responsibility to address these issues and our case studies have shown that they are aware of this and putting measures in place. This is clearly true for data protection or security issues. The medium-term issues that are less likely to find local resolutions still figure prominently, even though an individual organisation has less influence on how they can be addressed. Examples of this would be questions of unemployment, justice, or fairness. There was little reference to what we call long-term issues, which can partly be explained by the fact that the type of AI user organisations we investigated have very limited influence on how they are perceived and how they may be addressed.

Interpretative Differences on Ethical Issues

Despite general agreement on the terminology used to describe ethical issues, there are often important differences in interpretation and understanding. In the first ethics theme, control of data, the perceptions of privacy ranged from ‘not an issue’ to an issue that was being dealt with. Some of this arose from the question of informed consent and the GDPR. However, a reliance on legislation, such as GDPR, without full knowledge of the intricacies of its details (i.e. that informed consent is only one of several legal bases of lawful data processing), may give rise to a false sense of security over people’s perceived privacy. This was also linked to the issue of transparency (of processes dealing with data), which may be external to the organisation (do people outside understand how an organisation holds and processes their data), or internal (how well does the organisation understand the algorithms developed internally) and sometimes involve deliberate opacity (used in specific contexts where it is perceived as necessary, such as in monitoring political unrest and its possible consequences). Therefore, a clearer and more nuanced understanding of privacy and other ethical terms raised here might well be useful, albeit tricky to derive in a public setting (for an example of complications in defining privacy, see Macnish, 2018 ).

Some issues from the literature were not mentioned in the cases, such as warfare. This can easily be explained by our choice of case studies, none of which drew on work done in this area. It indicates that even a set of 10 case studies falls short of covering all issues.

A further empirical insight is in the category we called ‘role of organisations’, which covers trust and responsibility. Trust is a key term in the discussion of the ethics of AI, prominently highlighted by the focus on trustworthy AI by the EU’s High-Level Expert Group, among others. We put this into the ‘role of organisations’ category because our interaction with the case study respondents suggested that they felt it was part of the role of their organisations to foster trust and establish responsibilities. But we are open to the suggestion that these are concepts on a slightly different level that may provide the link between specific issues in applications and broader societal debate.

Next Steps: Addressing the Ethics of AI and Big Data

This paper is predominantly descriptive, and it aims to provide a theoretically sound and empirically rich account of ethical concerns in AI + BD. While we hope that it proves to be insightful it is only a first step in the broader journey towards addressing and resolving these issues. The categorisation suggested here gives an initial indication of which type of actor may be called upon to address which type of issue. The distinction between micro-, meso- and macro perspectives suggested by Haenlein and Kaplan ( 2019 ) resonates to some degree with our categorisation of issues.

This points to the question what can be done to address these ethical issues and by whom should it be done? We have not touched on this question in the theoretical or empirical part of the paper, but the question of mitigation is the motivating force behind much of the AI + BD ethics research. The purpose of understanding these ethical questions is to find ways of addressing them.

This calls for a more detailed investigation of the ethical nature of the issues described here. As indicated earlier, we did not begin with a specific ethical theoretical framework imposed onto the case studies, but did have some derived ethics concepts which we explored within the context of the cases and allowed others to emerge over the course of the interviews. One issue is the philosophical question whether the different ethical issues discussed here are of a similar or comparable nature and what characterises them as ethical issues. This is not only a philosophical question but also a practical one for policymakers and decision makers. We have alluded to the idea that privacy and data protection are ethical issues, but they also have strong legal implications and can also be human rights issues. It would therefore be beneficial to undertake a further analysis to investigate which of these ethical issues are already regulated and to what degree current regulation covers BD + AI, and how this varies across the various EU nations and beyond.

Another step could be to expand an investigation like the one presented here to cover the ethics of AI + BD debate with a focus on suggested resolutions and policies. This could be achieved by adopting the categorisation and structure presented here and extending it to the currently discussed option for addressing the ethical issues. These include individual and collective activities ranging from technical measures to measure bias in data or individual professional guidance to standardisation, legislation, the creation of a specific regulator and many more. It will be important to understand how these measures are conceptualised as well as which ones are already used to which effect. Any such future work, however, will need to be based on a sound understanding of the issues themselves, which this paper contributes to. The key contribution of the paper, namely the presentation of empirical findings from 10 case studies show in more detail how ethical issues play out in practice. While this work can and should be expanded by including an even broader variety of cases and could be supplemented by other empirical research methods, it marks an important step in the development of our understanding of these ethical issues. This should form a part of the broader societal debate about what these new technologies can and should be used for and how we can ensure that their consequences are beneficial for individuals and society.

Throughout the paper, XXX will be used to anonymise relevant text that may identify the authors, either through the project and/or publications resulting from the individual case studies. All case studies have been published individually. Several the XXX references in the findings refer to these individual publications which provide more detail on the cases than can be provided in this cross-case analysis.

The ethical issues that we discussed throughout the case studies refers to issues broadly construed as ethical issues, or issues that have ethical significance. While some issues may not be directly obvious how they are ethical issues, they may give rise to significant harm relevant to ethics. For example, accuracy of data may not explicitly be an ethical issue, if inaccurate data is used in algorithms, it may lead to discrimination, unfair bias, or harms to individuals.

Such as chat-bots, natural language processing AI, IoT data retrieval, predictive risk analysis, cybersecurity machine-learning, and large dataset exchanges.

https://ec.europa.eu/futurium/en/ai-alliance-consultation/guidelines/1 .

https://ec.europa.eu/digital-single-market/en/high-level-expert-group-artificial-intelligence .

The type of AI currently in vogue, as outlined earlier, is based on machine learning, typically employing artificial neural networks for big data analysis. This is typically seen as ‘narrow AI’ and it is not clear whether there is a way from narrow to general AI, even if one were to accept that achieving general AI is fundamentally possible.

The 16 social domains were: Banking and securities; Healthcare; Insurance; Retail and wholesale trade; Science; Education; Energy and utilities; Manufacturing and natural resources; Agriculture; Communications, media and entertainment; Transportation; Employee monitoring and administration; Government; Law enforcement and justice; Sustainable development; and Defence and national security.

This increased to 26 ethical issues following a group brainstorming session at the case study workshop.

The nine additional ethical issues from the initial 17 drafted by the project leader were: human rights, transparency, responsibility, ownership of data, algorithmic bias, integrity, human rights, human contact, and accuracy of data.

The additional ethical issues were access to BD + AI, accuracy of data, accuracy of recommendations, algorithmic bias, economic, human contact, human rights, integrity, ownership of data, responsibility, and transparency. Two of the initial ethical concerns were removed (inclusion of stakeholders and environmental impact). The issues raised concerning inclusion of stakeholders were deemed to be sufficiently included in access to BD + AI, and those relating to environmental impact were felt to be sufficiently covered by sustainability.

The three appendices attached in this paper comprise much of this case study protocol.

CS4 evaluated four organisations, but one of these organisations was also part of CS2 – Organisation 1. CS6 analysed two insurance organisations.

Starting out, we aimed to have both policy/ethics-focused experts within the organisation and individuals that could also speak with us about the technical aspects of the organisation’s BD + AI. However, this was often not possible, due to availability, organisations’ inability to free up resources (e.g. employee’s time) for interviews, or lack of designated experts in those areas.

For example, in CS1, CS6, and CS8.

For example, in CS2, CS3, CS4, CS5, CS6, and CS9.

As is discussed elsewhere in this paper, algorithms also hold the possibility of reinforcing our prejudices and biases or creating new ones entirely.

Accenture. (2016). Building digital trust: The role of data ethics in the digital age. Retrieved December 1, 2020 from https://www.accenture.com/t20160613T024441__w__/us-en/_acnmedia/PDF-22/Accenture-Data-Ethics-POV-WEB.pdf .

Accenture. (2017). Embracing artificial intelligence. Enabling strong and inclusive AI driven growth. Retrieved December 1, 2020 from https://www.accenture.com/t20170614T130615Z__w__/us-en/_acnmedia/Accenture/next-gen-5/event-g20-yea-summit/pdfs/Accenture-Intelligent-Economy.pdf .

Antoniou, J., & Andreou, A. (2019). Case study: The Internet of Things and Ethics. The Orbit Journal, 2 (2), 67.

Google Scholar  

Badri, A., Boudreau-Trudel, B., & Souissi, A. S. (2018). Occupational health and safety in the industry 4.0 era: A cause for major concern? Safety Science, 109, 403–411. https://doi.org/10.1016/j.ssci.2018.06.012

Article   Google Scholar  

Barolli, L., Takizawa, M., Xhafa, F., & Enokido, T. (ed.) (2019). Web, artificial intelligence and network applications. In Proceedings of the workshops of the 33rd international conference on advanced information networking and applications , Springer.

Barocas, S., & Selbst, A. D. (2016). Big data’s disparate impact. California Law Review, 104 (671), 671–732. https://doi.org/10.15779/Z38BG31

Baum, S. D. (2017). Reconciliation between factions focused on near-term and long-term artificial intelligence. AI Society, 2018 (33), 565–572.

Belliveau, K. M., Gray, L. E., & Wilson, R. J. (2019). Busting the Black Box: Big Data Employment and Privacy | IADC LAW. https://www.iadclaw.org/publications-news/defensecounseljournal/busting-the-black-box-big-data-employment-and-privacy/ . Accessed 10 May 2019.

Bossman, J. (2016). Top 9 ethical issues in artificial intelligence. World Economic Forum . https://www.weforum.org/agenda/2016/10/top-10-ethical-issues-in-artificial-intelligence/ . Accessed 10 May 2019.

Bostrom, N. (2016). Superintelligence: Paths . OUP Oxford.

Boyd, D., & Crawford, K. (2012). Critical questions for big data. Information, Communication and Society, 15 (5), 662–679. https://doi.org/10.1080/1369118X.2012.678878

Burrell, J. (2016). How the machine ‘thinks’: Understanding opacity in machine learning algorithms. Big Data and Society, 3 (1), 2053951715622512.

Bush, T., (2012). Authenticity in Research: Reliability, Validity and Triangulation. Chapter 6 in edited “Research Methods in Educational Leadership and Management”, SAGE Publications.

Calders, T., Kamiran, F., & Pechenizkiy, M. (2009). Building classifiers with independency constraints. In IEEE international conference data mining workshops , ICDMW’09, Miami, USA.

Chatfield, K., Iatridis, K., Stahl, B. C., & Paspallis, N. (2017). Innovating responsibly in ICT for ageing: Drivers, obstacles and implementation. Sustainability, 9 (6), 971. https://doi.org/10.3390/su9060971 .

Cohen, I. G., Amarasingham, R., Shah, A., et al. (2014). The legal and ethical concerns that arise from using complex predictive analytics in health care. Health Affairs, 33 (7), 1139–1147.

Couldry, N., & Powell, A. (2014). Big Data from the bottom up. Big Data and Society, 1 (2), 205395171453927. https://doi.org/10.1177/2053951714539277

Crawford, K., Gray, M. L., & Miltner, K. (2014). Big data| critiquing big data: Politics, ethics, epistemology | special section introduction. International Journal of Communication, 8, 10.

Cuquet, M., & Fensel, A. (2018). The societal impact of big data: A research roadmap for Europe. Technology in Society, 54, 74–86.

Danna, A., & Gandy, O. H., Jr. (2002). All that glitters is not gold: Digging beneath the surface of data mining. Journal of Business Ethics, 40 (4), 373–438.

European Convention for the Protection of HUman Rights and Fundamental Freedoms, pmbl., Nov. 4, 1950, 213 UNTS 221.

Herriott, E. R., & Firestone, W. (1983). Multisite qualitative policy research: Optimizing description and generalizability. Educational Researcher, 12, 14–19. https://doi.org/10.3102/0013189X012002014

Einav, L., & Levin, J. (2014). Economics in the age of big data. Science, 346 (6210), 1243089. https://doi.org/10.1126/science.1243089

Ferraggine, V. E., Doorn, J. H., & Rivera, L. C. (2009). Handbook of research on innovations in database technologies and applications: Current and future trends (pp. 1–1124). IGI Global.

Fothergill, B. T., Knight, W., Stahl, B. C., & Ulnicane, I. (2019). Responsible data governance of neuroscience big data. Frontiers in Neuroinformatics, 13 . https://doi.org/10.3389/fninf.2019.00028

Frey, C. B., & Osborne, M. A. (2017). The future of employment: How susceptible are jobs to computerisation? Technological Forecasting and Social Change, 114, 254–280. https://doi.org/10.1016/j.techfore.2016.08.019

Haenlein, M., & Kaplan, A. (2019). A brief history of artificial intelligence: On the past, present, and future of artificial intelligence. California Management Review, 61 (4), 5–14.

Harari, Y. N. (2017). Homo deus: A brief history of tomorrow (1st ed.). Vintage.

Book   Google Scholar  

ICO. (2017). Big data, artificial intelligence, machine learning and data protection. Retrieved December 1, 2020 from Information Commissioner’s Office website: https://iconewsblog.wordpress.com/2017/03/03/ai-machine-learning-and-personal-data/ .

Ioannidis, J. P. (2013). Informed consent, big data, and the oxymoron of research that is not research. The American Journal of Bioethics., 2, 15.

Jain, P., Gyanchandani, M., & Khare, N. (2016). Big data privacy: A technological perspective and review. Journal of Big Data, 3 (1), 25.

Janssen, M., & Kuk, G. (2016). The challenges and limits of big data algorithms in technocratic governance. Government Information Quarterly, 33 (3), 371–377. https://doi.org/10.1016/j.giq.2016.08.011

Jirotka, M., Grimpe, B., Stahl, B., Hartswood, M., & Eden, G. (2017). Responsible research and innovation in the digital age. Communications of the ACM, 60 (5), 62–68. https://doi.org/10.1145/3064940

Jiya, T. (2019). Ethical Implications Of Predictive Risk Intelligence. ORBIT Journal, 2 (2), 51.

Jiya, T. (2019). Ethical reflections of human brain research and smart information systems. The ORBIT Journal, 2 (2), 1–24.

Jobin, A., Ienca, M., & Vayena, E. (2019). The global landscape of AI ethics guidelines. Nature Machine Intelligence, 1 (9), 389–399. https://doi.org/10.1038/s42256-019-0088-2

Johnson, J. A. (2014). From open data to information justice. Ethics and Information Technology, 4 (16), 263–274.

Johnson, J. A. (2018). Open data, big data, and just data. In J. A. Johnson (Ed.), Toward information justice (pp. 23–49). Berlin: Springer.

Chapter   Google Scholar  

Kancevičienė, N. (2019). Insurance, smart information systems and ethics: a case study. The ORBIT Journal, 2 (2), 1–27.

Keeso, A. (2014). Big data and environmental sustainability: A conversation starter . https://www.google.com/search?rlz=1C1CHBF_nlNL796NL796&ei=YF3VXN3qCMLCwAKp4qjYBQ&q=Keeso+Big+Data+and+Environmental+Sustainability%3A+A+Conversation+Starter&oq=Keeso+Big+Data+and+Environmental+Sustainability%3A+A+Conversation+Starter&gs_l=psy-ab.3...15460.16163..16528...0.0..0.76.371.6......0....1..gws-wiz.......0i71j35i304i39j0i13i30.M_8nNbaL2E8 . Accessed 10 May 2019.

Kuriakose, F., & Iyer, D. (2018). Human Rights in the Big Data World (SSRN Scholarly Paper No. ID 3246969). Rochester, NY: Social Science Research Network. https://papers.ssrn.com/abstract=3246969 . Accessed 13 May 2019.

Kurzweil, R. (2006). The singularity is near . Gerald Duckworth & Co Ltd.

Latonero, M. (2018). Big data analytics and human rights. New Technologies for Human Rights Law and Practice. https://doi.org/10.1017/9781316838952.007

Lepri, B., Staiano, J., Sangokoya, D., Letouzé, E., & Oliver, N. (2017). The tyranny of data? the bright and dark sides of data-driven decision-making for social good. In Transparent data mining for big and small data (pp. 3–24). Springer.

Livingstone, D. (2015). Transhumanism: The history of a dangerous idea . CreateSpace Independent Publishing Platform.

Macnish, K. (2018). Government surveillance and why defining privacy matters in a post-snowden world. Journal of Applied Philosophy, 35 (2), 417–432.

Macnish, K., & Inguanzo, A. (2019). Case study-customer relation management, smart information systems and ethics. The ORBIT Journal, 2 (2), 1–24.

Macnish, K., Inguanzo, A. F., & Kirichenko, A. (2019). Smart information systems in cybersecurity. ORBIT Journal, 2 (2), 15.

Mai, J. E. (2016). Big data privacy: The datafication of personal information. The Information Society, 32 (3), 192–199.

Manson, N. C., & O’Neill, O. (2007). Rethinking informed consent in bioethics . Cambridge University Press.

Mittelstadt, B. D., Allo, P., Taddeo, M., Wachter, S., & Floridi, L. (2016). The ethics of algorithms: Mapping the debate. Big Data and Society, 3 (2), 2053951716679679.

Meeker, Q. W., & , Hong, Y. . (2014). Reliability Meets big data: Opportunities and challenges. Quality Engineering, 26 (1), 102–116.

Newman, N. (2013). The costs of lost privacy: Consumer harm and rising economic inequality in the age of google (SSRN Scholarly Paper No. ID 2310146). Rochester: Social Science Research Network. https://papers.ssrn.com/abstract=2310146 . Accessed 10 May 2019.

O’Neil, C. (2016). Weapons of math destruction: How big data increases inequality and threatens democracy . Crown Publishers.

Panch, T., Mattie, H., & Atun, R. (2019). Artificial intelligence and algorithmic bias: implications for health systems. Journal of global health, 9 (2).

Pellé, S., & Reber, B. (2015). Responsible innovation in the light of moral responsibility. Journal on Chain and Network Science, 15 (2), 107–117. https://doi.org/10.3920/JCNS2014.x017

Portmess, L., & Tower, S. (2015). Data barns, ambient intelligence and cloud computing: The tacit epistemology and linguistic representation of Big Data. Ethics and Information Technology, 17 (1), 1–9. https://doi.org/10.1007/s10676-014-9357-2

Ryan, M. (2019). Ethics of public use of AI and big data. ORBIT Journal, 2 (2), 15.

Ryan, M. (2019). Ethics of using AI and big data in agriculture: The case of a large agriculture multinational. The ORBIT Journal, 2 (2), 1–27.

Ryan, M., & Gregory, A. (2019). Ethics of using smart city AI and big data: The case of four large European cities. The ORBIT Journal, 2 (2), 1–36.

Sadowski, J., & Pasquale, F. A. (2015). The spectrum of control: A social theory of the smart city. First Monday, 20 (7), 16.

Schradie, J. (2017). Big data is too small: Research implications of class inequality for online data collection. In D. June & P. Andrea (Eds.), Media and class: TV, film and digital culture . Abingdon: Taylor and Francis.

Taylor, L. (2017). ‘What is data justice? The case for connecting digital rights and freedoms globally’ In Big data and society (pp. 1–14). https://doi.org/10.1177/2053951717736335 .

Tene, O., & Polonetsky, J. (2012). Big data for all: Privacy and user control in the age of analytics. The Northwestern Journal of Technology and Intellectual Property, 11, 10.

Tene, O., & Polonetsky, J. (2013). A theory of creepy: technology, privacy and shifting social norms. Yale JL and Technology, 16, 59.

Van Dijck, J., & Poell, T. (2013). Understanding social media logic. Media and Communication, 1 (1), 2–14.

Voinea, C., & Uszkai, R. (n.d.). An assessement of algorithmic accountability methods .

Walsham, G. (1995). Interpretive case studies in IS research: nature and method. European Journal of Information Systems, 4 (2), 74–81.

Wallace, N., & Castro, D. (2018) The Impact of the EU’s New Data Protection Regulation on AI, Centre for Data Innovation .

Walsham, G. (1995). Interpretive case-studies in IS research-nature and method. European Journal of Information Systems, 4 (2), 74–81.

Walsham, G. (2006). Doing interpretive research. European Journal of Information Systems, 15 (3), 320–330.

Wheeler, G. (2016). Machine epistemology and big data. In L. McIntyre & A. Rosenburg (Eds.), Routledge Companion to Philosophy of Social Science . Routledge.

Whittlestone, J., Nyrup, R., Alexandrova, A., Dihal, K., & Cave, S. (2019). Ethical and societal implications of algorithms, data, and artificial intelligence: A roadmap for research. https://www.nuffieldfoundation.org/sites/default/files/files/Ethical-and-Societal-Implications-of-Data-and-AI-report-Nuffield-Foundat.pdf .

Wolf, B. (2015). Burkhardt Wolf: Big data, small freedom? / Radical Philosophy. Radical Philosophy . https://www.radicalphilosophy.com/commentary/big-data-small-freedom . Accessed 13 May 2019.

Yin, R. K. (2014). Case study research: Design and methods (5th ed.). SAGE.

Yin, R. K. (2015). Qualitative research from start to finish . Guilford Publications.

Zwitter, A. (2014). Big data ethics. Big Data and Society, 1 (2), 51.

Zuboff, S. (2015). Big other: Surveillance capitalism and the prospects of an information civilization (April 4, 2015). Journal of Information Technology, 2015 (30), 75–89. https://doi.org/10.1057/jit.2015.5

Download references

Acknowledgements

This SHERPA Project has received funding from the European Union’s Horizon 2020 Framework Programme for Research and Innovation under the Specific Grant Agreement No. 786641. The author(s) acknowledge the contribution of the consortium to the development and design of the case study approach.

Author information

Authors and affiliations.

Wageningen Economic Research, Wageningen University and Research, Wageningen, The Netherlands

UCLan Cyprus, Larnaka, Cyprus

Josephina Antoniou

De Montford University, Leicester, UK

Laurence Brooks & Bernd Stahl

Northampton University, Northampton, UK

Tilimbe Jiya

The University of Twente, Enschede, The Netherlands

Kevin Macnish

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Mark Ryan .

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendix 1: Desk Research Questions

Number Research Question.

In which sector is the organisation located (e.g. industry, government, NGO, etc.)?

What is the name of the organisation?

What is the geographic scope of the organisation?

What is the name of the interviewee?

What is the interviewee’s role within the organisation?

Appendix 2: Interview Research Questions

No Research Question.

What involvement has the interviewee had with BD + AI within the organisation?

What type of BD + AI is the organisation using? (e.g. IBM Watson, Google Deepmind)

What is the field of application of the BD + AI (e.g. administration, healthcare, retail)

Does the BD + AI work as intended or are there problems with its operation?

What are the innovative elements introduced by the BD + AI (e.g. what has the technology enabled within the organisation?)

What is the level of maturity of the BD + AI ? (i.e. has the technology been used for long at the organisation? Is it a recent development or an established approach?)

How does the BD + AI interact with other technologies within the organisation?

What are the parameters/inputs used to inform the BD + AI ? (e.g. which sorts of data are input, how is the data understood within the algorithm?). Does the BD + AI collect and/or use data which identifies or can be used to identify a living person (personal data)?. Does the BD + AI collect personal data without the consent of the person to whom those data relate?

What are the principles informing the algorithm used in the BD + AI (e.g. does the algorithm assume that people walk in similar ways, does it assume that loitering involves not moving outside a particular radius in a particular time frame?). Does the BD + AI classify people into groups? If so, how are these groups determined? Does the BD + AI identify abnormal behaviour? If so, what is abnormal behaviour to the BD + AI ?

Are there policies in place governing the use of the BD + AI ?

How transparent is the technology to administrators within the organisation, to users within the organisation?

Who are the stakeholders in the organisation?

What has been the impact of the BD + AI on stakeholders?

How transparent is the technology to people outside the organisation?

Are those stakeholders engaged with the BD + AI ? (e.g. are those affected aware of the BD + AI, do they have any say in its operation?). If so, what is the nature of this engagement? (focus groups, feedback, etc.)

In what way are stakeholders impacted by the BD + AI ? (e.g. what is the societal impact: are there issues of inequality, fairness, safety, filter bubbles, etc.?)

What are the costs of using the BD + AI to stakeholders? (e.g. potential loss of privacy, loss of potential to sell information, potential loss of reputation)

What is the expected longevity of this impact? (e.g. is this expected to be temporary or long-term?)

Are those stakeholders engaged with the BD + AI ? (e.g. are those affected aware of the BD + AI, do they have any say in its operation?)

If so, what is the nature of this engagement? (focus groups, feedback, etc.)

Appendix 3: Checklist of Ethical Issues

Rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Ryan, M., Antoniou, J., Brooks, L. et al. Research and Practice of AI Ethics: A Case Study Approach Juxtaposing Academic Discourse with Organisational Reality. Sci Eng Ethics 27 , 16 (2021). https://doi.org/10.1007/s11948-021-00293-x

Download citation

Received : 26 August 2019

Accepted : 10 February 2021

Published : 08 March 2021

DOI : https://doi.org/10.1007/s11948-021-00293-x

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Smart information systems
• Big data analytics
• Artificial intelligence ethics
• Multiple-case study analysis
• Philosophy of technology
• Find a journal
• Publish with us
• Track your research

• Explore AI by Industry PLUS
• Consumer goods
• Heavy industry
• Natural resources
• Professional services
• Transportation
• AI Best Practice Guides PLUS
• AI White Paper Library PLUS
• AI Business Process Explorer PLUS
• Enterprise AI Newsletter
• Emerj Plus Research
• AI in Business Podcast
• The AI Consulting Podcast
• AI in Financial Services Podcast
• Precisely – Building Trust in Data
• Shift Technology – How Insurers are Using AI
• Uniphore – The Future of Banking CX in APAC
• Uniphore – The Economic Impact of Conversational AI and Automation
• Uniphore – The Future of Complaints Management
• Uniphore – Conversational AI in Banking

AI Case Studies Articles and Reports

Explore in-depth case studies of real artificial intelligence applications in business, including applications in marketing, finance, security, and other sectors.

One of Australasia’s Largest Banks Unifies Their AI Efforts

The following is a case study for Emerj's AI Opportunity Landscape research. To learn more about how we help companies develop winning AI strategies and identify the highest-ROI applications, watch the two-minute video summary of our AI Opportunity Landscape research. Problem The bank had many scattered AI projects, but struggled with:

Online Education Company Improves Customer Support with Autosuggestion of Macros

Technology Provider: DigitalGenius is an artificial intelligence solutions provider for customer service operations.

Automotive Repair Equipment OEM uses AI to Monetize Repair Service Data

Tech Provider: Predii - a company which specializes in building AI platforms designed for repair and maintenance services in the industrial equipment and automotive sector

The Department of Homeland Security Uses AI-Enhanced Entity Resolution for its Global Travel Assessment System (GTAS)

Tech Provider: TAMR - a company that automates the organization of a company’s enterprise-wide data through a machine learning-driven plus human-guided solution

Smart Home Services Provider Automates Report Creation with AI and Customer Data

Technology Provider: Automated Insights is a provider of natural language generation platform, Wordsmith, that converts big data into narratives or summary understandable by humans

User Engagement Company Improves Lead Quality Using Predictive Analytics

Technology Provider: Ignite Technologies (which acquired Infer, a Mountain View, CA-based marketing optimization and machine learning company).

Smart Home Services Provider Uses Natural Language Generation to Create Highly Personalized Website Copy

Technology Provider: Automated Insights is a provider of a natural language generation platform, Wordsmith, that converts big data into narratives or summaries understandable by humans.

Microsoft Gets the Pulse of Customer Sentiment with Natural Language Processing

Tech Provider:  Lexalytics - A natural language processing and text analytics software and service provider. Client Company: Microsoft - A multinational technology company which develops, licenses, sells and supports software, hardware, and services. Industry:  Consumer goods Function: Business intelligence and analytics, Process automation Tech Provider Contact Person: Charles Le Gros

• Market Reasearch and Advisory
• AI Presentations and Keynotes
• Emerj Plus Membership
• AI In Business Podcast
• AI In Finance Services Podcast
• Subscribe to our AI Newsletter
• Advertise with us
• Terms and Conditions
• Refund and Cancellation Policy
• Privacy Policy

AI in cybersecurity an introduction and case studies

An introduction to AI in cybersecurity with real-world case studies in a Fortune 500 organization and a government agency

Despite all the recent advances in artificial intelligence and machine learning (AI/ML) applied to a vast array of application areas and use cases, success in AI in cybersecurity remains elusive. The key component to building AI/ML applications is training data, which traditionally must be manually labeled by subject matter experts (SMEs) in order to train ML models to perform a given task. Organizations seeking to develop custom AI/ML applications for cyber use cases face numerous challenges, each of which is addressed by Snorkel Flow:

• Availability of training data : Unlike other domains, where open-source training datasets are often readily available, datasets featuring real-world cyber events within enterprise networks are rarely released, stifling the development of AI/ML applications. Organizations are left with the daunting task of creating custom labeled datasets required to train ML models.
• Truly massive scale:  The amount of network traffic generated on a daily basis by enterprises of every size is almost inconceivably large. Relying on teams of cyber analysts to manually review and label all of this data by hand in order to train ML models is a complete non-starter.
• No two networks are alike : Each enterprise network and policy environment is unique, requiring custom ML training datasets specific to each network’s unique characteristics. The varying definitions of normal versus malicious traffic, and the distinction between authorized and prohibited activities, render “one-size-fits-all” training data and models largely ineffective.
• Dynamic threat environment:  Cyber adversaries are creative and adaptive, constantly evolving their tactics, techniques, and procedures to circumvent security measures. Static solutions that don’t allow network defenders to respond to constantly-changing threats will always be one step behind.

Snorkel Flow  enables organizations to quickly and easily build their own labeled ML training datasets in-house, on their own network data and relying on their own cyber analysts’ knowledge. This allows them to train custom AI applications specifically suited to the unique nature of their networks and policy environments, and adapt these applications rapidly to accommodate network/policy changes or shifting adversary tactics. All of this can be achieved without any sensitive data leaving the organization within a fully governable and auditable workflow. 

In this series of articles, we highlight two major case studies in which Snorkel AI customers overcame these challenges to develop AI/ML applications for cyber use cases with Snorkel Flow:

• A Fortune 500 telecommunications provider uses Snorkel Flow to classify encrypted network data flows
• A U.S. Government agency uses Snorkel Flow for application classification and network attack detection

AI in cybersecurity case study: Fortune 500 telco uses Snorkel Flow to classify encrypted network data flows

A globally-recognized Fortune 500 telecom company used Snorkel Flow to classify encrypted network data flows into their associated application categories. The customer’s data science team faced a number of challenges for this task, including:

• Their past experience of  labeling network traffic data by hand was too slow, noisy, and expensive , requiring precious time from network data experts.
• Their existing probe used a static set of rules based on a fixed set of Service Name Indicators (SNIs), leading to a  brittle solution that was difficult to adapt .
• Their previous approach also struggled to accommodate changing data distributions, such as responding to network trouble tickets or alarms.
• Multiple tools were required to cover the end-to-end machine learning pipeline, from data exploration and labeling, to model training and analysis.

Task description: Classifying network data flows

To evaluate Snorkel Flow’s ability to  accelerate the development of ML models and AI applications for network data use cases , this customer compared a new solution they developed in Snorkel Flow against an existing solution based on a fully-supervised ML model trained on 178,000 ground-truth data examples. The Snorkel Flow solution was also compared against a baseline model trained on a subset of 2,000 examples from this ground-truth dataset. The goal was to see whether Snorkel Flow could enable this organization to produce a solution capable of  replacing the fully-supervised ML in a fraction of the time required to manually label a ground-truth training dataset.

In this data, individual flows contained categorical and text features like source/destination port and IP address, SNIs, and forward/backward packets. Flows were then preprocessed to include statistical features, such as forward/backward inter-arrival time (IAT) statistics, flow bytes per second, and flow packets per second. Users of varying skill levels combined a number of different strategies for this effort, including creating no-code labeling functions with Snorkel Flow’s built-in network data visualization tools, writing code-based labeling functions using Snorkel Flow’s Python SDK, and leveraging Snorkel Flow’s ability to auto-generate labeling functions based on self-training and semi-supervised methods.

Results with Snorkel Flow

From an initial subset of only 2,000 ground-truth labeled examples, this telecom customer used Snorkel Flow to produce an additional 198,000 programmatically-labeled examples. A model trained in Snorkel Flow on this 200,000-example training dataset was  26.2% more accurate than a baseline model  trained on the subset of 2,000 examples, and  within 0.2% accuracy of a fully-supervised model  trained on all 178,000 ground-truth examples. For aspects of the data that change over time, such as SNIs (which are prone to drift in production settings), the Snorkel Flow model was compared with a static rules-based solution and the baseline model described above. Incredibly,  the Snorkel Flow model was 77.3% more accurate than the rules-based approach and also outperformed the baseline model by nearly 10% . An additional experiment was performed to test Snorkel Flow’s ability to adapt to changing distributions in the data, like the proportions between application types in the network traffic. Here, the Snorkel Flow model beat the baseline model by over 20% and managed to slightly outperform a fully-supervised model as well.

Ultimately, using Snorkel Flow enabled this customer to:

• Deliver high-accuracy ML models for a network data application quickly , without being slowed down by an extensive hand-labeling process.
• Develop adaptable solutions that provide a marked improvement over brittle, rules-based approaches.
• Build applications that are  robust to network data drift , by maintaining consistently-high performance scores even when data distributions change.
• Help network data experts and data scientists to work together, using  first-class network visualization and data processing utilities  in a unified platform.

Encouraged by these impressive results on their first Snorkel Flow project, this customer is now developing a new application focused on anomaly detection for IMS network equipment metrics. This effort will  identify real-time anomalies over time-series data , starting with the number of call attempts per second.

AI in cybersecurity case study: U.S. Government agency uses Snorkel Flow for application classification and network attack detection

A major AI center of excellence within the U.S. government selected Snorkel Flow to evaluate  programmatic labeling  as a means of accelerating the development of AI/ML applications for multiple cyber use cases. For this, ML models created in Snorkel Flow were compared with models trained on pre-labeled datasets. This group recognized hand-labeling as inadequate for their purposes for a variety of reasons:

• Sensitive data cannot be shared externally for crowdsourced or outsourced labeling.  Internal access is also frequently restricted to only those with a valid need-to-know, making the task of labeling ML training data even more difficult.
• Labeling data by hand leads to ineffective workforce utilization.  Experienced cyber SMEs are in high demand and short supply. There simply aren’t enough available to hand-label millions (or tens of millions) of individual data points.
• Labeling data isn’t a “one and done” task.  ML training data must be frequently updated to adapt to changes in real-world inputs or shifts in organizational objectives, which only magnifies the scalability issue.
• ML models trained on hand-labeled data lack explainability.  For an important policy decision or military action, it’s essential to articulate precisely why a specific choice was made. The need for  trustworthy AI  extends this to AI/ML.

Task description: Application type classification

The dataset for this task consisted of network packets collected from different applications, described using 50 data features and containing over 2,700,000 total records. The feature set included a mix of packet statistics (e.g., packet count, length of traffic, etc.), along with source/destination IP addresses and ports. Destination IP is very useful for predicting application type, but generally not advisable for use in training an ML model because applications often change the destination IPs used over time, and a model trained on this feature would not be able to adapt to such changes. Snorkel Flow provided a way to use the destination IP to quickly label training data with labeling functions, while preventing the resulting ML model from using it for training in lieu of more reliable packet statistics. In Snorkel Flow, we refer to this as a  “non-servable” feature : a feature that can be used for labeling the data, but should not be relied upon as a feature for model training or prediction.

That is, as shown in the image below, destination IP was incorporated as a non-servable feature for the labeling functions (LFs) used to label the training data, alongside packet statistics and port information as servable features. However, at inference time, the trained model only relies on the servable features (i.e., packet statistics and port information),  excluding destination IP as a non-servable feature.

Using Snorkel Flow, the customer’s team of data scientists and cyber SMEs programmatically labeled nearly 280,000 records using just 6 labeling functions.  All the labeling functions had extremely high precision, with four achieving 100% precision, one hitting 99%, and another reaching 96% precision, and each with coverage proportional to the class distribution in the original dataset. An ML model trained in Snorkel Flow on this programmatically-labeled dataset was compared with a baseline model trained on 2,000 pre-labeled examples (representing a typical hand-labeled dataset for this kind of task).  The Snorkel Flow model trained on the programmatically-labeled dataset outperformed the baseline by 7.4 points on an F1 scale.  Additionally, the built-in targeted error analysis tools in Snorkel Flow were leveraged to discover and address several errors in the original dataset for this task. For example, a number of DNS traffic records were improperly labeled (by the source) as various different application types, rather than being labeled as DNS traffic itself. Snorkel Flow made it easy to quickly filter the dataset to identify these examples, leading the customer’s cyber SMEs to determine that all DNS traffic should be treated the same, rather than being grouped inconsistently into traffic examples from different applications.  Overall, based on the error analysis performed in Snorkel Flow and the data errors detected, 41% of the original dataset was relabeled.

Task description: Network attack detection (Port scan vs. Benign traffic)

For this second task, a dataset of ~30,000 network packets with 85 features (e.g., packet length, IAT, etc.) was used to characterize traffic as either benign or part of a  port scan attack . Instead of relying on interactive SME input in Snorkel Flow, here is an  existing network attack ontology was directly transformed into labeling functions.

With just two labeling functions derived from the ontology,  100% of the original dataset was programmatically labeled in Snorkel Flow  (i.e., all 30,000 examples). An ML model trained on the programmatically-labeled dataset achieved 88.1% accuracy.

Based on these initial successes, this federal customer is now exploring further work  using Snorkel Flow to enable other mission-critical applications for AI in cybersecurity , and also extending the deployment of Snorkel Flow to non-cyber areas like  financial analysis .

For organizations building AI applications using programmatic labeling and the  data-centric AI  approach pioneered by Snorkel and enabled by Snorkel Flow,  solving all of these challenges becomes possible.  To learn more about Snorkel and how we can help your organization accelerate the development of AI solutions for cyber use cases, or to  request a demo , follow us on  Twitter ,  Linkedin ,  Facebook , or  Instagram . or  contact us  to learn more. 

Join the community

You might also like.

Recommended reading

Build high-quality ai radically faster.

Case Study on Application of Artificial Intelligence to Oral English Teaching in Vocational Colleges

Ieee account.

• Change Username/Password
• Update Address

Purchase Details

• Payment Options
• Order History
• View Purchased Documents

Profile Information

• Communications Preferences
• Profession and Education
• Technical Interests
• US & Canada: +1 800 678 4333
• Worldwide: +1 732 981 0060
• Contact & Support
• About IEEE Xplore
• Accessibility
• Terms of Use
• Nondiscrimination Policy
• Privacy & Opting Out of Cookies

A not-for-profit organization, IEEE is the world's largest technical professional organization dedicated to advancing technology for the benefit of humanity. © Copyright 2024 IEEE - All rights reserved. Use of this web site signifies your agreement to the terms and conditions.

Artificial Intelligence Case Study Topics

Looking for artificial intelligence case study topics? Explore real-life examples and learn how AI is transforming industries like healthcare, finance, manuf...

Artificial Intelligence Case Study Topics: Unleashing the Power of AI

Artificial Intelligence (AI) has emerged as one of the most transformative technologies in recent times, revolutionizing industries and reshaping the way we live and work. With its ability to analyze vast amounts of data, learn from patterns, and make autonomous decisions, AI has the potential to solve complex problems and unlock new possibilities. One of the key drivers of AI advancements is the utilization of case studies, which provide real-world examples of AI applications and their impact.

Introduction to AI Case Studies

Case studies serve as invaluable resources in understanding the practical applications of AI. They offer insights into how AI technologies are implemented, the challenges faced, and the outcomes achieved. By examining successful AI case studies, we can gain a deeper understanding of the potential of AI and how it can be harnessed to drive innovation and improve various aspects of our lives.

The Importance of AI Case Studies

AI case studies play a pivotal role in showcasing the capabilities of AI systems and their potential impact. These studies enable researchers, developers, and businesses to learn from past experiences, identify best practices, and avoid potential pitfalls. By studying successful AI case studies, decision-makers can make informed choices when implementing AI solutions, ensuring maximum efficiency and effectiveness.

Purpose of the Blog Post

The purpose of this blog post is to provide an in-depth exploration of artificial intelligence case study topics. We will delve into various industries and domains where AI has made significant strides, examining real-life examples and their impact. By the end of this comprehensive guide, you will have a clear understanding of the potential applications of AI across different sectors and gain insights into how these case studies have transformed industries.

Overview of Artificial Intelligence Case Studies

Before we dive into specific case studies, let's first establish a foundational understanding of AI case studies. These case studies involve the application of AI technologies to address a specific problem or challenge. They provide a detailed account of how AI systems were developed, implemented, and the outcomes achieved.

AI case studies offer a multifaceted perspective, encompassing various industries, including healthcare, finance, manufacturing, customer service, and transportation. Each case study presents a unique set of challenges and opportunities, highlighting the versatility and adaptability of AI in different contexts.

Real-life Examples of Successful AI Case Studies

To truly grasp the potential of AI, it is essential to explore real-life examples of successful AI case studies. These pioneering projects have showcased the transformative power of AI, pushing the boundaries of what was once thought possible. Let's take a glimpse into some notable AI case studies:

1. Google DeepMind's AlphaGo

In 2016, Google's DeepMind developed AlphaGo, an AI system that defeated the world champion Go player, Lee Sedol. This groundbreaking achievement highlighted the ability of AI to master complex strategic games that were previously considered beyond the reach of machines. AlphaGo's success demonstrated the potential of AI in problem-solving and decision-making in complex scenarios.

2. IBM Watson's Jeopardy! Victory

IBM's Watson showcased its cognitive capabilities by competing against human champions on the popular quiz show, Jeopardy! in 2011. Watson's ability to understand and process natural language, coupled with its vast knowledge base, enabled it to outperform the human contestants. This case study demonstrated the potential of AI in understanding and analyzing unstructured data, paving the way for advancements in natural language processing.

3. Tesla's Autopilot System

Tesla's Autopilot system utilizes AI algorithms and sensors to enable semi-autonomous driving. By analyzing real-time data from cameras, radar, and ultrasonic sensors, the Autopilot system can detect and respond to road conditions, other vehicles, and pedestrians. This case study showcases the potential of AI in the transportation industry, revolutionizing the concept of self-driving cars.

4. Amazon's Recommendation Engine

Amazon's recommendation engine is powered by AI algorithms that analyze customer preferences, purchase history, and browsing behavior to provide personalized product recommendations. This case study demonstrates how AI can enhance the customer experience by delivering targeted suggestions, improving sales, and fostering customer loyalty.

These real-life examples are just the tip of the iceberg when it comes to AI case studies. They illustrate the diverse range of industries and domains where AI has made significant contributions, showcasing the potential for innovation and transformation.

In the next section, we will explore the process of selecting artificial intelligence case study topics, considering various factors and identifying the most relevant and impactful areas of study. Stay tuned for an in-depth analysis of AI case studies in healthcare, finance, manufacturing, customer service, and transportation.

Note: In the following sections, we will explore each case study topic in greater detail, analyzing the problem at hand, the AI solution implemented, and the results and impact achieved.

Artificial intelligence (AI) case studies provide valuable insights into the practical applications and impact of AI technologies. These case studies offer a glimpse into the real-world implementation of AI systems, showcasing their capabilities, successes, and challenges. By examining these case studies, we can gain a deeper understanding of the potential of AI and its ability to transform various industries.

Explanation of AI Case Studies

AI case studies involve the application of AI technologies to solve specific problems or challenges within a given context. These studies provide detailed accounts of how AI systems were developed, implemented, and the outcomes achieved. By analyzing the methodologies and approaches used in these case studies, researchers, developers, and businesses can learn from past experiences and gain insights into the best practices for implementing AI solutions.

AI case studies often involve the utilization of machine learning algorithms, natural language processing, computer vision, robotics, and other AI techniques. They can range from small-scale projects to large-scale deployments, depending on the complexity of the problem being addressed.

Benefits of AI Case Studies

AI case studies offer numerous benefits for both researchers and practitioners in the field of AI. Here are some key advantages:

Insights into Implementation : Case studies offer insights into the practical implementation of AI systems. They provide details on the data collection process, model training, algorithm selection, and optimization techniques employed. This information can guide future AI projects and help avoid common pitfalls.

Benchmarking and Comparison : Case studies allow for benchmarking and comparison of different AI approaches. By examining multiple case studies within a specific domain, researchers can identify the strengths and weaknesses of various AI techniques, leading to advancements and improvements in the field.

Inspiration for Innovation : AI case studies can inspire new ideas and innovative solutions. By understanding the challenges faced in previous case studies and the methods used to overcome them, researchers can build upon existing knowledge and push the boundaries of AI capabilities.

To truly comprehend the power and potential of AI, it is essential to explore real-life examples of successful AI case studies. These examples highlight the impact that AI can have across various domains. Let's take a closer look at some notable AI case studies:

Google DeepMind's AlphaGo : AlphaGo, developed by Google DeepMind, made headlines in 2016 when it defeated the world champion Go player, Lee Sedol. This case study demonstrated the ability of AI to master complex strategic games and showcased the potential for AI in decision-making and problem-solving.

IBM Watson's Jeopardy! Victory : In 2011, IBM's Watson competed against human champions on the quiz show Jeopardy! and emerged victorious. Watson's success demonstrated the power of AI in natural language processing and understanding unstructured data.

Tesla's Autopilot System : Tesla's Autopilot system utilizes AI algorithms and sensors to enable semi-autonomous driving. This case study showcases the potential of AI in the transportation industry, revolutionizing the concept of self-driving cars.

Amazon's Recommendation Engine : Amazon's recommendation engine utilizes AI to analyze customer preferences and provide personalized product recommendations. This case study highlights how AI can enhance the customer experience and drive sales through targeted suggestions.

These real-life examples illustrate the diverse range of industries and domains where AI has made significant contributions. They serve as inspiration and provide valuable insights into the potential of AI technologies.

Choosing Artificial Intelligence Case Study Topics

When exploring the world of artificial intelligence case studies, it is essential to select the right topics that align with current AI trends and have the potential for significant impact. In this section, we will discuss the factors to consider when choosing case study topics and identify some promising areas for exploration.

Factors to Consider

Relevance to Current AI Trends : Selecting case study topics that align with current AI trends ensures that you are exploring areas of research and development that are actively advancing. Staying up-to-date with the latest advancements in AI will provide you with a better understanding of the challenges and opportunities in the field.

Availability of Data : Data availability is crucial for successful AI case studies. Consider topics where relevant and high-quality data is accessible. Adequate data sets are essential for training AI models effectively and obtaining reliable results.

Ethical Considerations : Ethical considerations should be an integral part of AI case study topic selection. It is important to choose topics that adhere to ethical guidelines and prioritize fairness, transparency, and accountability. Avoid topics that raise concerns regarding privacy, bias, or potential harm to individuals or society.

Identifying Potential Case Study Topics

Now, let's explore some potential case study topics in various industries where AI has shown promising applications:

Healthcare and Medical Diagnostics : AI has the potential to revolutionize healthcare by improving diagnostics, predicting disease outcomes, and enabling personalized treatment plans. Some potential case study topics in this domain include:

AI in Early Cancer Detection: Explore how AI algorithms can analyze medical imaging data to detect and diagnose cancer at an early stage, leading to improved patient outcomes.

AI in Medical Imaging Analysis: Investigate how AI can assist radiologists in analyzing medical images, such as X-rays, MRIs, and CT scans, to improve accuracy and speed in diagnosis.

Financial Services and Fraud Detection : AI offers significant potential in the finance industry, particularly in fraud detection and prevention. Some potential case study topics in this domain include:

AI in Fraud Detection for Banks: Examine how AI algorithms can analyze transaction data and detect fraudulent activities in real-time, enhancing security and minimizing financial losses.

AI in Credit Card Fraud Detection: Explore how AI can analyze patterns and anomalies in credit card transactions to identify and prevent fraudulent activities, ensuring the safety of customers' financial information.

Manufacturing and Process Optimization : AI can optimize manufacturing processes, improve efficiency, and reduce costs. Some potential case study topics in this domain include:

AI in Predictive Maintenance: Investigate how AI can analyze sensor data to predict machinery failures and schedule maintenance proactively, minimizing downtime and optimizing production.

AI in Supply Chain Optimization: Explore how AI algorithms can optimize supply chain operations by predicting demand, optimizing inventory levels, and improving logistics, leading to cost savings and improved customer satisfaction.

Customer Service and Chatbots : AI-powered chatbots have revolutionized customer service by providing instant responses and personalized experiences. Some potential case study topics in this domain include:

AI-powered Chatbots in E-commerce: Examine how AI-powered chatbots can enhance customer engagement, provide personalized product recommendations, and streamline the online shopping experience.

AI in Virtual Assistants for Customer Support: Explore how AI-based virtual assistants can handle customer inquiries, resolve issues, and provide 24/7 support, improving customer satisfaction and reducing support costs.

Transportation and Autonomous Vehicles : AI plays a critical role in the development of autonomous vehicles and traffic management systems. Some potential case study topics in this domain include:

AI in Self-Driving Cars: Investigate how AI algorithms enable autonomous vehicles to perceive the environment, make real-time decisions, and navigate safely on the roads.

AI in Traffic Management Systems: Explore how AI can optimize traffic flow, reduce congestion, and improve transportation efficiency by analyzing real-time traffic data and implementing intelligent control systems.

By considering these factors and exploring potential case study topics in various industries, you can select areas that align with your interests and have the potential to contribute to the advancement of AI technologies.

Deep Dive into Selected Artificial Intelligence Case Study Topics

In this section, we will delve deeper into selected artificial intelligence case study topics across various industries. By examining these case studies, we can gain a comprehensive understanding of the problem at hand, the AI solutions implemented, and the results and impact achieved.

Healthcare and Medical Diagnostics

Case Study: AI in Early Cancer Detection

Overview of the Problem: Early detection of cancer is crucial for successful treatment and improved patient outcomes. However, it can be challenging for healthcare professionals to accurately detect cancer at its early stages due to the complexity of medical imaging data and the potential for human error.

AI Solution and Implementation: In this case study, AI algorithms were developed and trained using large datasets of medical imaging data, including mammograms, CT scans, or MRIs. These algorithms utilize deep learning techniques to analyze and interpret the images, identifying potential cancerous cells or tumors. By comparing the patterns in the images to an extensive database of known cancer cases, the AI system can provide accurate early detection of cancer.

Results and Impact: The implementation of AI in early cancer detection has shown promising results. The AI system can analyze medical images with high accuracy, often outperforming human radiologists in detecting cancer at its early stages. Early detection allows for timely intervention, leading to improved treatment outcomes and increased survival rates for patients.

Case Study: AI in Medical Imaging Analysis

Overview of the Problem: Medical imaging, such as X-rays, MRIs, and CT scans, plays a crucial role in diagnosing and monitoring various medical conditions. However, the interpretation of these images can be time-consuming, subjective, and prone to human error.

AI Solution and Implementation: In this case study, AI algorithms were developed and trained using large datasets of labeled medical imaging data. These algorithms leverage deep learning techniques, such as convolutional neural networks (CNNs), to analyze and interpret the images. The AI system can identify anomalies, highlight potential abnormalities, and provide quantitative measurements to assist radiologists in making accurate diagnoses.

Results and Impact: The implementation of AI in medical imaging analysis has shown significant potential in improving diagnostic accuracy and efficiency. The AI system can assist radiologists in identifying subtle abnormalities that may be missed by the human eye, leading to early detection of diseases and improved patient care. Additionally, AI can help reduce the burden on radiologists by automating certain tasks, allowing them to focus on more complex cases.

Financial Services and Fraud Detection

Case Study: AI in Fraud Detection for Banks

Overview of the Problem: Fraudulent activities, such as identity theft and unauthorized transactions, pose significant challenges for banks and financial institutions. Traditional rule-based fraud detection systems often struggle to keep up with evolving fraud techniques and patterns.

AI Solution and Implementation: In this case study, AI algorithms were developed to analyze large volumes of transactional data in real-time. These algorithms utilize machine learning techniques, including anomaly detection and pattern recognition, to identify suspicious activities that deviate from normal patterns. By continuously learning from new data, the AI system can adapt and evolve to detect new and emerging fraud patterns.

Results and Impact: The implementation of AI in fraud detection for banks has led to improved fraud prevention and detection rates. The AI system can analyze vast amounts of transactional data quickly and accurately, flagging potentially fraudulent activities in real-time. By minimizing false positives and identifying fraudulent transactions promptly, banks can mitigate financial losses and protect their customers' assets.

Case Study: AI in Credit Card Fraud Detection

Overview of the Problem: Credit card fraud is a significant concern for both financial institutions and cardholders. Detecting fraudulent credit card transactions is challenging due to the large volume of transactions and the need for real-time analysis.

AI Solution and Implementation: In this case study, AI algorithms were developed to analyze credit card transaction data, including transaction amounts, merchant information, and cardholder behavior. These algorithms utilize machine learning techniques, such as supervised and unsupervised learning, to identify patterns and anomalies indicative of fraudulent activities. The AI system can learn from historical data to improve its fraud detection capabilities over time.

Results and Impact: The implementation of AI in credit card fraud detection has proven to be highly effective in reducing fraudulent activities. The AI system can quickly analyze transactions, identify suspicious patterns, and flag potentially fraudulent transactions for further investigation. By minimizing false positives and accurately detecting fraud, financial institutions can protect their customers and maintain trust in the credit card ecosystem.

In the next section, we will explore case studies in manufacturing and process optimization, showcasing how AI can enhance efficiency and streamline operations.

In this section, we will explore case studies in the domain of manufacturing and process optimization. These examples highlight how artificial intelligence (AI) can enhance efficiency, reduce costs, and streamline operations in manufacturing industries.

Manufacturing and Process Optimization

Case Study: AI in Predictive Maintenance

Overview of the Problem: Unplanned equipment failures and unexpected downtime can significantly impact manufacturing operations, leading to production delays and increased costs. Traditional maintenance strategies, such as reactive or preventive maintenance, may not effectively address the challenges of equipment failure prediction and maintenance scheduling.

AI Solution and Implementation: In this case study, AI algorithms were implemented to perform predictive maintenance. The algorithms utilize machine learning techniques, such as supervised learning and anomaly detection, to analyze sensor data from machines and predict potential failures. By continuously monitoring the health and performance of equipment, the AI system can identify early warning signs of impending failures and schedule maintenance proactively.

Results and Impact: The implementation of AI in predictive maintenance has proven to be highly beneficial for manufacturing industries. By detecting potential equipment failures in advance, companies can plan maintenance activities more efficiently, minimizing downtime and reducing costs associated with unscheduled repairs. This proactive approach to maintenance helps optimize production schedules and ensures smooth operations.

Case Study: AI in Supply Chain Optimization

Overview of the Problem: Supply chain management involves complex processes, including demand forecasting, inventory management, and logistics planning. Optimizing these processes is crucial for reducing costs, improving customer satisfaction, and increasing operational efficiency.

AI Solution and Implementation: In this case study, AI algorithms were utilized to optimize supply chain operations. The algorithms leverage machine learning techniques, such as demand forecasting, inventory optimization, and route optimization, to analyze historical and real-time data. By considering factors such as customer demand, lead times, transportation costs, and inventory levels, the AI system can generate optimal plans and recommendations for procurement, production, and distribution.

Results and Impact: The implementation of AI in supply chain optimization has led to significant improvements in efficiency and cost reduction. By accurately forecasting demand and optimizing inventory levels, companies can minimize stockouts and excess inventory, leading to reduced carrying costs and improved cash flow. AI-powered route optimization helps streamline logistics operations, optimizing delivery schedules and reducing transportation costs. These advancements in supply chain optimization ultimately lead to improved customer satisfaction through faster and more reliable deliveries.

These case studies highlight the potential impact of AI in manufacturing and process optimization. By leveraging AI technologies, companies can achieve greater efficiency, reduced costs, and improved operational effectiveness. In the next section, we will explore case studies in the domain of customer service and chatbots, showcasing how AI can enhance customer experiences and support interactions.

In this section, we will explore case studies in the domain of customer service and chatbots. These examples highlight how artificial intelligence (AI) can enhance customer experiences, streamline support interactions, and improve overall customer satisfaction.

Customer Service and Chatbots

Case Study: AI-powered Chatbots in E-commerce

Overview of the Problem: With the rise of e-commerce, providing personalized and timely customer support has become a crucial aspect of the online shopping experience. However, scaling customer service to meet the growing demands of a large customer base can be challenging and costly.

AI Solution and Implementation: In this case study, AI-powered chatbots were implemented to handle customer inquiries and provide support in e-commerce platforms. These chatbots utilize natural language processing (NLP) and machine learning algorithms to understand and respond to customer queries. They can provide instant and personalized responses, offer product recommendations based on customer preferences, and assist with order tracking and returns.

Results and Impact: The implementation of AI-powered chatbots in e-commerce has significantly improved customer experiences and operational efficiency. Chatbots provide instant responses, reducing customer wait times and ensuring 24/7 availability for support inquiries. By offering personalized product recommendations, chatbots can enhance the shopping experience and increase sales conversion rates. Additionally, chatbots can handle routine inquiries, freeing up human agents to focus on more complex customer issues, ultimately improving overall customer satisfaction.

Case Study: AI in Virtual Assistants for Customer Support

Overview of the Problem: Customer support departments often face high call volumes and long wait times, leading to customer frustration and decreased satisfaction. Providing timely and effective support to customers is critical for maintaining brand loyalty and positive customer experiences.

AI Solution and Implementation: In this case study, AI-powered virtual assistants were implemented to handle customer support interactions. These virtual assistants utilize AI technologies such as natural language processing, sentiment analysis, and knowledge graph systems. They can understand customer inquiries, provide accurate and personalized responses, and escalate complex issues to human agents when necessary. Virtual assistants continuously learn from customer interactions, improving their responses and problem-solving abilities over time.

Results and Impact: The implementation of AI-powered virtual assistants in customer support has proven to be highly effective in improving response times and customer satisfaction. Virtual assistants can provide instant support, reducing wait times and enabling customers to receive assistance at their convenience. By accurately understanding customer inquiries and providing relevant information, virtual assistants can resolve issues quickly and efficiently. This results in improved customer experiences, reduced support costs, and increased customer loyalty.

These case studies illustrate the potential of AI in enhancing customer service and support interactions. By leveraging AI-powered chatbots and virtual assistants, businesses can provide timely, personalized, and efficient support to their customers, resulting in improved customer satisfaction and loyalty. In the next section, we will explore case studies in the domain of transportation and autonomous vehicles, showcasing how AI is revolutionizing the way we travel and manage traffic.

In this section, we will explore case studies in the domain of transportation and autonomous vehicles. These examples highlight how artificial intelligence (AI) is revolutionizing the way we travel and manage traffic.

Transportation and Autonomous Vehicles

Case Study: AI in Self-Driving Cars

Overview of the Problem: Self-driving cars have the potential to transform the transportation industry by reducing accidents, improving traffic flow, and enhancing overall mobility. However, developing autonomous vehicles that can navigate safely and make real-time decisions in complex traffic scenarios is a significant challenge.

AI Solution and Implementation: In this case study, AI algorithms are utilized to power self-driving cars. These algorithms leverage a combination of computer vision, sensor fusion, machine learning, and decision-making models to perceive the environment, interpret traffic signs, detect obstacles, and make real-time driving decisions. By continuously analyzing sensor data and learning from past experiences, self-driving cars can navigate autonomously while adhering to traffic rules and ensuring passenger safety.

Results and Impact: The implementation of AI in self-driving cars has the potential to revolutionize transportation. Autonomous vehicles can reduce human errors and improve road safety by eliminating the risks associated with human distraction, fatigue, and impaired driving. Additionally, self-driving cars have the potential to optimize traffic flow, reduce congestion, and increase overall transportation efficiency, leading to reduced travel times and fuel consumption.

Case Study: AI in Traffic Management Systems

Overview of the Problem: Managing traffic flow in urban areas is a complex task that requires real-time analysis of traffic patterns, congestion, and accidents. Traditional traffic management systems often struggle to handle the dynamic nature of traffic and effectively optimize traffic flow.

AI Solution and Implementation: In this case study, AI algorithms are used to enhance traffic management systems. These algorithms leverage machine learning techniques and real-time data analysis to predict traffic congestion, optimize signal timings, and suggest alternative routes. By analyzing historical and real-time traffic data, the AI system can make intelligent decisions to improve traffic flow, reduce congestion, and minimize travel times.

Results and Impact: The implementation of AI in traffic management systems has shown significant potential in improving transportation efficiency. By optimizing signal timings based on real-time traffic conditions, AI can reduce congestion and ensure a smoother flow of vehicles. AI algorithms can also provide real-time traffic updates to drivers, enabling them to make informed decisions about alternative routes, further reducing travel times and improving overall traffic management.

These case studies highlight how AI is transforming the transportation industry. From self-driving cars to intelligent traffic management systems, AI technologies have the potential to revolutionize the way we travel, making transportation safer, more efficient, and environmentally friendly.

In this comprehensive guide, we have explored various artificial intelligence case study topics across different industries. We have witnessed the power of AI in healthcare, finance, manufacturing, customer service, and transportation. By examining real-life examples and understanding the problem-solving capabilities of AI, we have gained insights into the potential of this transformative technology.

AI case studies provide invaluable lessons and inspire innovation in the field of artificial intelligence. They offer opportunities for learning, benchmarking, and improving AI systems. By studying successful case studies, researchers, developers, and businesses can harness the power of AI to drive advancements, solve complex problems, and improve various aspects of our lives.

As AI continues to evolve, it is crucial to stay updated with the latest trends, research, and case studies. The potential of AI is immense, and by exploring and sharing knowledge, we can collectively shape a future where AI-driven solutions enhance our lives in remarkable ways.

Adrian Kennedy

• Open access
• Published: 18 December 2023

Challenges and opportunities of AI in inclusive education: a case study of data-enhanced active reading in Japan

• Yuko Toyokawa   ORCID: orcid.org/0000-0003-2386-3303 1 ,
• Izumi Horikoshi 2 ,
• Rwitajit Majumdar 2 , 3 &
• Hiroaki Ogata 2  

Smart Learning Environments volume  10 , Article number:  67 ( 2023 ) Cite this article

4110 Accesses

2 Citations

2 Altmetric

Metrics details

In inclusive education, students with different needs learn in the same context. With the advancement of artificial intelligence (AI) technologies, it is expected that they will contribute further to an inclusive learning environment that meets the individual needs of diverse learners. However, in Japan, we did not find any studies exploring current needs in an actual special needs context. In this study, we used the learning and evidence analysis framework (LEAF) as a learning analytics-enhanced learning environment and employed Active Reading as an example learning task to investigate the challenges and possibilities of applying AI to inclusive education in the future. Two students who attended a resource room formed the context. We investigated learning logs in the LEAF system while each student executed a given learning task. We detected specific learning behaviors from the logs and explored the challenges and future potential of learning with AI technology, considering human involvement in orchestrating inclusive educational practices.

Introduction

Efforts are underway to promote the realization of inclusive education and the widespread development of inclusive environments in which all children can learn together irrespective of their disabilities, cultural backgrounds, or socioeconomic status (UNESCO, 2009 ). Inclusive education in Japan primarily focuses on learners with disabilities and aims to enable them to actively participate in and contribute to society independently in an inclusive manner (MEXT, 2012 ). In general, not only in Japan, but also in many other countries, students with mild disabilities, such as those with developmental disorders or disabilities (DD), study alongside non-disabled learners in the same learning environment in regular classes in inclusive education. In diverse but constrained learning contexts with different types of learners, teachers have difficulty orchestrating multiple flows of information and tasks (Dillenbourg, 2013 ). Although there are many different types of educational practices within inclusive education, special education (SE) approaches can be used to meet and support the unique learning needs of learners with special needs in a learning environment (Bryant et al., 2019 ).

In regular classes, all learners engage in learning at the same pace, but students with learning difficulties (LD), who are said to be less efficient at processing information, tend to have trouble catching up in class compared with other students (Gersten et al., 2001 ). This may cause depression, poor academic performance, and low self-esteem (Peterson et al., 2001 ; Rose, 2019 ). For such learners, resource rooms or pullout programs can provide extra support outside regular classes (Bryant et al., 2019 ). A resource room under inclusive education in Japan is an independent remedial class in which learners with a relatively mild disability, or those who tend to demonstrate some difficulties, leave their regular classes and receive support according to their needs (MEXT, 2020 ). In the learning context, Toyokawa and her colleagues observed that students in a resource room in Japan implemented daily learning activities with a digital e-book reader called BookRoll in the learning and evidence analysis framework (LEAF) with learning analytics (LA) technology and found the possibility of detecting their stumbling points and strengths in their learning logs (Toyokawa et al., 2022 ). A large amount of data can be accumulated from daily learning using LEAF. However, the utilization of LA technology such as LEAF for learners with special needs has not been researched extensively in an inclusive Japanese learning environment. More than 30 years ago, Yin argued about the future-oriented investigation of new technology, including using artificial intelligence (AI) in SE (Yin & Moore, 1987 ). Research on inclusive education using AI has been rapidly gaining attention worldwide (Kazimzade et al., 2019 ; Salas-Pilco et al., 2022 ). However, just as Kazimzade and her colleagues mentioned the lack of exchange between AI and disability research in their book chapter (Kazimzade et al., 2019 ), the lack of progress in the context of special needs is also the case in Japan. Therefore, we propose integrating LA and AI technology to effectively orchestrate learning for learners with special needs in inclusive education. Focusing on literacy skills that underlie all aspects of learning and daily life and bearing in mind the importance of reading, we selected active reading (AR) in an LA-enhanced learning environment as one task and investigated the challenges and possibilities of AI integration.

The remainder of this paper is organized as follows. In the second section, an overview of inclusive education in Japan, LA-enhanced learning environments, and AI in inclusive education is presented. In the third section, the research objectives and a question are stated, and then the LEAF components are introduced as a learning environment for this study, followed by participants and learning tasks. Data collection and interpretation are then described. The following section presents the findings of the case study to answer the research question. In the Discussion section, possible solutions for learning with AI are discussed along with limitations for future research. Finally, the implications and contributions of the study are highlighted.

Literature review

Special education in inclusive education in japan.

In inclusive educational environments, students study together in the same class, regardless of their difficulties. Inclusive education is defined as education in which students with disabilities have access to the standard curriculum in a general education classroom (Bryant et al., 2019 ). In the Japanese inclusive context, students with relatively mild DD [e.g., autism, low vision, speech impairment, attention deficit hyperactivity disorder (ADHD), and LD] attend the same classes as students without DD. In Japan, the number of students with DD is increasing. According to a report from the Ministry of Education in Japan (MEXT), the number was approximately 600,000 in 2012 and 800,000 in 2022, or approximately two to three students with DD out of every thirty students in one class (MEXT, 2022a ). For such learners, a resource room or pullout program is available, and which provides extra support outside of regular classes upon request (Bryant et al., 2019 ). The support system differs depending on needs, but attending a resource room is the primary form of receiving additional support at school for learners with DD in the current inclusive education system in Japan. The Japanese resource room is an independent supplementary class in which learners with relatively mild disabilities or those who tend to show some difficulty leaving regular classes receive special support equivalent to self-reliance activities according to their needs (MEXT, 2020 ). Learners with various difficulties can receive support tailored to their individual needs, such as social or communication skills training and academic support, such as reading, writing, and math. In this respect, resource rooms can be said to be a part of SE, in which learners with difficulties can receive support based on their needs. SE is an approach designed to meet the unique learning needs of individuals with disabilities, such as students with different learning, behavioral, social communication, and basic functional needs (Bryant et al., 2019 ). Currently, the resource room service is provided at elementary schools, junior high schools, and high schools in Japan, but the situation is that there are students who need support but are left unattended for reasons such as a lack of instructors (MEXT, 2022b ).

Information and communication technology (ICT) in education is said to be progressing in Japan, but the penetration rate lags far behind that of other countries when looking at the average Program for International Student Assessment (PISA) of the Organisation for Economic Co-operation and Development (OECD) (National Institute for Educational Policy Research, 2022 ). Research on the use of ICT in SE in Japan has primarily focused on alternative and assistive technologies and teaching materials (Kinoshita et al., 2023 ; Kumagai & Nagai, 2022 ). While research on technical assistance has garnered considerable attention in the literature, there is a notable gap in research pertaining to special needs in inclusive education from the lens of LA. This gap is especially pronounced in the context of Japan, where the utilization of learning log data and AI technology for this purpose remains unexplored.

Learning analytics and support for learners with special needs

Using e-learning tools such as ICT, it is possible to collect and accumulate learning log data that record the learning process. LA, which is research on the contribution of learning logs to learning and educational activities, has attracted attention. LA is research aimed at improving and enhancing teaching and learning by analyzing and visualizing accumulated log data and providing feedback based on the visualization through daily learning activities (Bodily & Verbert, 2017 ; Siemens & Baker, 2012 ). Using the LA learning system LEAF, Toyokawa and her colleagues traced students' handwriting from their interaction performance in the daily learning of students attending a resource room in an elementary school in Japan to investigate their learning performance and difficulties (Toyokawa et al., 2022 ). In this study, they successfully visualized and observed learning behaviors such as students’ learning difficulties using penstroke analysis. This study demonstrated the possibility of using log data to assist learners with special needs and support teachers. To cite two overseas examples, first, a pilot study was conducted in which a learning game for cognitively impaired people was developed and learning behavior was observed from interaction and performance data using LA (Buzzi et al., 2016 ). The learning game allows for assigning and monitoring tasks remotely, encouraging learning according to individual needs, and analyzing the results obtained from learning. The second example is an attempt to provide support by opening a learner model using LA and detecting reading difficulties, such as learning style and cognitive traits, from the demographic submodel and reading profile (Mejia et al., 2016 ). This study underscores the importance that learners are aware of their own learning styles and cognitive limitations. All three cases sought to support learners with special needs and teachers using LA. It is expected that the LA-enhanced learning environment will further improve learning and education with AI technology in the future; however, in Japan, LA research to support learning has not yet become popular in SE. Furthermore, limited research has provided AI-based support for the unique requirements of inclusive education.

AI in inclusive education

AI has the capacity to harness learners' behavioral data, ultimately delivering personalized and tailored educational services to cater to individual needs, as suggested by Margetts and Dorobantu ( 2019 ). AI also aids in making more accurate predictions and planning learning. According to the same study, some local governments in the UK are already using predictive analytics to anticipate future needs in areas such as SE and children’s social services. This prediction can also be applied to identify students who are considered to be “at risk” (Cano & Leonard, 2019 ; Slowík et al., 2021 ). Such warning systems are already in use in the United States, New Zealand, and Canada.

AI has also had a significant impact on Japanese society. Although educational big data have been accumulated through the use of ICT and machine learning, compared to other countries, it is obvious that in Japan, AI technology in the educational field lags behind the national level. Kazimzade and her colleagues argue that most of today's adaptive education systems rarely consider diversity and that it is necessary to create heterogeneous data sets to train AI in inclusive learning environments to replicate our diverse societies (Kazimzade et al., 2019 ). This lack of heterogeneous datasets is particularly evident in the context of SE in inclusive education in Japan. In this respect, this research is one of the few to focus on learning support using AI for minority learners who need special support in Japan. In this study, we investigated how to support learners with special needs in inclusive education using AI technology. The research methods and experimental procedures are described in the next section.

Research objective

Given the need to understand how AI-driven approaches can realize future SE in inclusive education in the Japanese context, we conducted a case study to explore the current needs, challenges, and opportunities of implementing AI.

What are the challenges and opportunities of AI-driven services for active reading of learners with special needs in inclusive education?

Case studies have gained considerable acceptance as valid research methods in a wide range of fields. In particular, Yin’s case study is said to be reliable for connecting the underlying theory and practice (Zainal, 2007 ). A case study enables us to understand behavioral states from the perspective of learners and subjects, which is said to be useful in explaining the complexity of real learning situations in detail (Zainal, 2007 ). Research on learning in SE is a large field; however, only a limited number of individuals can be selected as research subjects. It is valuable to accumulate data obtained from daily learning in a natural way, and we consider this experiment “a unique way of observing natural phenomena present in a series of data,” as defined by Yin’s case study (Zainal, 2007 ). Next, we present the LEAF system as a reading learning environment and workflow that were utilized to investigate the challenges and opportunities of AI applications.

LEAF system and its components used in a case study

We propose the use of the LEAF as an LA-enhanced AR learning environment for inclusive education. LEAF is a learning environment framework that includes BookRoll, an e-learning material browsing system that allows learners to view digital learning materials anytime and anywhere, and a group of LA dashboard modules (LogPalette) that analyze and visualize the logs learned using BookRoll (Ogata et al., 2018 ). BookRoll includes reading-facilitating functions such as markers that can be used for highlighting and memos that can be added as annotations. Learners can choose input methods such as keyboards, direct handwriting using a stylus pen, and text conversion from voice input. Learning logs, such as the contents of memos, portions highlighted with markers and their content, number of operations, and viewing time, are accumulated in the Learning Record Store and analyzed and visualized in LogPalette. Figure 1 illustrates the LEAF framework with BookRoll and LogPalette interfaces.

Examples of the BookRoll interface, the LA dashboard, and the pen stroke analysis interface in the LEAF framework

Participants and study context

The participants were two twelve-year-old boys (boys 1 and 2). Boy 1 attended a resource room for six years to receive social communication training and had received special support before entering elementary school. Boy 2 was diagnosed with autism and attended a resource room for six years. He received special support before entering elementary school. Resource rooms are for students with relatively mild difficulties, and many who attend these rooms have not been diagnosed with disabilities. The decision on whether one is to receive special support in a resource room is made by the school principal, following an appropriate understanding of the actual situation and a discussion with the school committee (MEXT, 2020 ). Therefore, in this study, no details on the difficulty level were available for each child. The participants were asked to perform AR at home with their mothers. Written informed consent was obtained from the guardians of the students. First, the flow of learning activities was explained to the students and their mothers. Then, all four AR activities for Boy 1 which lasted about for one hour, and three AR activities for Boy 2, which lasted approximately one and a half hours were observed by a researcher. They chose a device to use, either a PC or an iPad, and chose an input method, such as using a keyboard for typing or a stylus pen for handwriting. In Japan, under the Global and Innovation Gateway for All (GIGA) school initiatives, each student is provided with one device. Both students had no problems operating PCs and/or tablets and typing on keyboards at home by themselves. We asked them to work on their reading on their favorite device with the intention of doing it in a stress-free environment as much as possible. A case study was conducted on two students using BookRoll. We explain the reading-learning activities and AR procedure in the next section.

AR learning task

The two boys read the same four reading materials. They read individual stories using BookRoll. The reading process followed the AR process, which was performed using BookRoll in a past study (Toyokawa et al., 2023 ). First, in the pre-reading phase, participants were asked to have an image of the story they were going to read by looking at the page (title, pictures, etc.) and write their predictions in a memo. They were then asked to formulate questions based on their thoughts. Questions were also asked to be recorded in a memo. Each story contained questions on comprehension. While they read the text, they read the story as they looked for answers while marking the answers to the question with a marker directly on BookRoll. In the post-reading phase, participants reflected on their reading and wrote the content of the story in their own words. One week later, they were asked to recall the story and write about what they had remembered. We additionally communicated the AR learning process to both the resource room teacher of Boy 1 and the mother of Boy 2 with the dashboard, engaging in a reflective discussion and receiving their valuable feedback. The objectives and activities for each phase of the AR activities are explained in Table 1 .

Data collection and analysis

The time spent reading and operation logs were investigated to understand each participant's AR process. First, the time taken for each reading task was extracted from the time logs, including the time taken to complete one AR session, the time taken to make a prediction and questions in the pre-reading phase, the time taken to answer questions while reading and marking the answers with a marker, and the time taken to write down what was understood in the post-reading phase (Table  2 ). The objective was to check whether there were any characteristics of reading difficulty, such as taking too long to read, input, and output. Then, behaviors such as frequent page flipping, noticeable writing, erasing, and highlighting actions were visualized as a plot (Fig.  2 ) to understand if we could detect any reading difficulties in the logs and at what stage of AR intervention was required. In order to investigate the reading behaviors, logged actions such as OPEN, MEMO, HANDWRITING MEMO, MARKER, NAVIGATION, TIMER, BOOKMARK, and CLOSE were extracted and analyzed, whose descriptions and interpretations of action logs are listed in Table 3 . After the AR learning, as part of the experiment, we asked the resource room teacher of Boy 1 and the mother of Boy 2 to see each student's AR process and the visualized logs, and received their impressions and comments.

Log visualization of the AR behavior among the three students

Analysis of the participants’ time logs

First, we investigated the learning behavioral patterns found in the learning logs regarding the time spent on each AR task. What the two of them have in common is that it took a considerably long time to write a summary (paraphrasing in their own words) after reading. Boy 2 took three times as long as Boy 1 to do the same. The average time spent on summaries for Boy 1 was (m = 6.28 for 3 summaries), which is approximately 76% of the total average AR activity for Boy 1. The average time spent on summaries for Boy 2 was (m = 22.37 for 2 summaries), which is approximately 96% of total AR activity. A summary of the time spent on the AR tasks is presented in Table  2 .

Analysis of the participants’ operation logs

We then attempted to visualize the AR performance of the two participants from the operation log, which is depicted in the plots in Fig. 2 . Overall, we confirmed that the participants progressed to AR according to the following AR procedure: pre-, while-, and post-reading phases. What we could clearly observe from the plots was that during the first AR activity, Boy 2 with LD noticeably wrote and erased his handwriting, and during the second AR activity, he frequently flipped pages, touched the timer, and wrote and erased his memos. The third AR seemed to proceed smoothly without any extra action; however, the fourth AR was not conducted.

Analysis of the stakeholders’ interviews

In general, learners check and reflect on their own learning processes, but this time, we asked the resource room teacher and the mother of Boy 2 to observe the data, reflect on the learning, and give us their comments. Their comments were as follows:

The teacher told us that all learning with paper is stored in a file and shared with the parents during the interviews, which are conducted twice a year. Students' data are always collected and reported to schools. She said that it would be nice if they could accumulate and share what they had learned using (electronic) tools. She also mentioned that parents need to (and want to) know what their children are doing in school. Boy 2’s mother said that her son cannot get rid of his obsession with things he cannot do. Due to this, he cannot move on to the next task, and as a result, he cannot complete the task. She told us that she made posters so that her son could visually check the tasks, but he now makes his own to-do list daily and keeps it in his school bag. She said that being able to see what he is doing through his learning logs helps her understand and accept how he is doing in school.

In this section, we discuss the findings from the case study, which can serve as evidence for identifying future challenges and possibilities related to the application of AI technology to SE in inclusive education.

Erratic learning engagement of students with LD in different phases of the learning activities and during technology usage

Learners have different time engagements and approaches to the same learning task. In this study on AR activities, Boy 2 required more time than Boy 1 (Table 2 ). The observations demonstrated that Boy 2 approached each activity carefully. He paid particular attention to the order in which things appeared in the story and the flow of AR itself. He was initially overly focused then lost concentration, gave up on the way, and could not complete the tasks. It was also found from the observations that it took time for him to write his summary with a stylus pen on an iPad for the first AR activity. He appeared unfamiliar with the act of writing directly on the iPad screen with a pen, but enjoyed using a new tool. He did not use handwriting during the second AR session but used the keyboard with which he was already familiar. From the logs and observations, we understood that it might be time consuming for some learners to perform knowledge output activities, such as writing what they have understood.

Regarding technology use (Fig. 2 ), Boy 1 had relatively fewer extra actions in the logs besides AR activities, whereas Boy 2 had a greater number of extra actions that demonstrated fixation behavior on ICT features. For example, several operation logs were detected in terms of handwritten memos, such as ADD and DELETE, during the first task. In the second reading task, several additional page movements and timer operations were observed (Fig. 2 ). In the third task, it was observed that AR was completed without additional operations on the logs. However, it was observed that he lost concentration and motivation. Consequently, he was unable to start or complete the fourth task. We also found that learners may end up concentrating on things other than learning, such as using e-learning features, such as timers. These pedagogical challenges must be addressed when creating learning designs for students with special needs.

Varied understanding of stakeholders about data-driven learning

In this study, we faced difficulty obtaining the consent of the guardians for the experiments because AR was not the type of learning support that they had originally requested. Some parents did not consent to the collection of their children’s learning data. During the interviews, we found that there was still a lack of awareness about data-driven learning, such as how BookRoll is actually used for learning and how logs are used to support learning. However, it was also clear that the teacher and the mother were looking forward to the possibility of employing data-driven learning and sharing learning processes effectively using technology.

In this section, we first discuss the limitations of the current study and then address the possibilities and challenges of AI-driven special needs learning in inclusive education.

Limitations and solutions for the sample size

One of the limitations of the current study is its sample size, as there were only two subjects. In resource rooms in Japan, class activities are usually offered by one teacher to either individual students or small groups for a limited time. Therefore, only a limited number of students can receive support each day. In addition, not all schools in Japan have resource rooms. Hence, it was difficult to recruit a large number of participants for this study, even if subjects were collected from multiple schools. Additionally, some parents were not willing to participate in the research and did not consent, making recruiting subjects a major challenge. Thus, it may be difficult to apply and generalize the results of the current study to a broader context. In addition, the small sample size may suggest the possibility of bias in the data analysis. To minimize this possibility, we used log data from the participants' learning process and attempted to visualize the data in plots instead of collecting data from conventional sources such as surveys, tests, and observations. Two researchers performed the confirmation and interpretation of the logs. The results confirmed that differences in the reading process between the two participants, such as differences in how they approached AR and how they used the tools, were interpreted in the same way. Learning evaluations and decision-making regarding whether to provide students with support have often been made based on the evaluation of learners' artifacts, observations, survey results, communication among stakeholders, and subjective measures such as teachers' perceptions or parents' intentions, which may lead to biased judgments or unnecessary support. Although these assessment methods remain essential, by being able to clearly show artifacts and the learning process through log visualization, not only researchers, but also school administrators, teachers, and parents can objectively judge a child's learning progress and make decisions about support provision.

Improving learning design for continuous learning

As mentioned in the existing literature, the majority of research and experiments on reading-based learning typically conclude at the end of the study period, often failing to foster lasting reading habits among learners (Gersten et al., 2001 ). We must acknowledge that there was a need to repeatedly conduct AR activities over time in this study as well. Additionally, it is difficult for learners who have difficulty concentrating to continue learning if they are not satisfied with their learning activities. Designing learning activities to suit learners’ needs and preferences is necessary for learning satisfaction and continuation (Salas-Pilco et al., 2022 ). The AR procedure employed in this study was segmented into three phases. However, taking learners’ attention spans into account, it is imperative to focus on AI applications that offer precise, individualized guidance and feedback for more effective interventions. AI assists learners in learning at their own pace outside the classroom and school. Learners can then use the dashboard to monitor the learning process and learn to reflect and understand so that they can develop and improve their cognitive and metacognitive skills. Learning activities and pedagogical approaches should be improved so that learners with special needs can continue learning independently even after the experimental period ends.

Implications for usability enhancement of the LEAF platform for SE

Existing dashboards in LEAF have an environment in which general students can reflect. However, current AR-D in LEAF may or may not be suitable for learners with special needs. Therefore, we consider updating and improving the performance and content of the functions and systems regarding the concept of the Universal Design of Learning (UDL) (Rose & Meyer, 2002 ). This is because system affordances and dashboard designs can significantly impact perception, behavior, and acquisition. Improvements in the usability, accessibility, and reliability of the system are often indicated in past studies (Buzzi et al., 2016 ; Mejia et al., 2016 ). Improving the system and developing an AI-driven LA dashboard based on real data should be considered so that all learners, including students with special needs and their stakeholders, can easily manage their learning and reflect on it, which will help mitigate learners’ difficulties.

Log data-driven solutions and potentials of AI for AR

In this study, we observed variations in the time needed for AR and the approach adopted for the same learning task among different learners. Students with LD have been found to process information inefficiently and not to understand appropriate reading strategies, which can lead to unexpected learning failures in comprehension and decoding (Gersten et al., 2001 ). For such learners, it is essential to present the steps of “what has been achieved” and “what needs to be done” explicitly and offer cues to help them complete the task and progress to the next step (Gersten et al., 2001 ). In today's data-driven learning environments, such as LEAF, it is possible to notify learners of task completion and reward them to boost their self-esteem and motivation to read and learn. The utilization of log data may lead to more efficient learning. Further, AI complements learners' previous knowledge and skills. For example, it would be possible to use natural language generation to support reading-learning by navigating the contents and the flow of reading activities in an easy-to-understand manner using both text and audio. First, we demonstrate each phase of a potential AI-driven AR approach in the future based on the results of a case study.

[Pre-reading phase]

Although learners with LD are good at many things, they are said to fall behind other students in reading comprehension because of difficulties like making predictions and having limited imagination and cognitive biases (Randi et al., 2010 ). However, such students can be instructed to improve their reading comprehension by using pre-reading strategies that activate their attention and prior knowledge (Gersten et al., 2001 ). AR uses information such as visual and auditory aids to help learners create an image of what they are about to read before (or even while) reading. However, for students who are struggling with reading, AI automatically measures the time required, the length, and the difficulty of a text, integrates it with information from the accumulated learner's data such as their reading speed, weakness, and preferences, and assists them in the reading process. For example, for students who have difficulty imagining textual information, AI generates and provides visual information to make visualization easier. For learners who have difficulty following the order of learning activities, AI can aide learners with audio or textual guides or ask them what they want next to guide their learning. It may also display filters to help students choose what to do next or use past data to calculate the time required for each learner to learn and intervene to complete a task at the appropriate time. In addition, it may activate the learners' existing knowledge by guiding them to vocabulary quizzes and chapters related to the reading content, and provide information relevant to the content they are about to read. In this way, when learners become stuck and cannot predict or create an image of the story during the pre-reading phase, AI may intervene to stimulate their previous knowledge and offer assistance, such as by providing an advanced organizer framework (Idol-Maestas, 1985 ) to guide them on what to do next.

[While-reading phase]

There are various types of reading difficulties given as examples, such as difficulty with concentrating on one thing, following procedures, completing task thing through to the end, reading information from a text alone, and inability to empathize with the emotions and viewpoints of the characters, or just simply taking too long to read (Randi et al., 2010 ; Ryan, 2007 ). AI can offer cues to help learners maintain focus on their reading objectives and assist them in identifying corrective actions when necessary steps are not completed. When unnecessary actions are detected, AI can redirect learners' attention towards the task at hand. AI may thus enable learners with special needs to work on AR learning alone, which was said to be difficult for them (Gersten et al., 2001 ). At the current stage, we developed and tested a text recommender in the LEAF system that automatically recommends reading materials based on the logs from markers used for vocabulary during AR. In the future, AI will recommend reading materials that match learners' levels and preferences based on the outcomes from the AR activities, such as different stroke orders, selecting wrong characters, spelling errors, and frequently used words and content stored in memos. AI will assist in making connections with previously read materials and helping students consolidate and develop what they have read by recommending chapters to review and reading materials to work on next. Moreover, AI may act as a reading agent or invite peers and teachers as intermediaries for reciprocal teaching interventions and mutual guidance that improves reading comprehension through communication with others. In this way, AI may provide opportunities for learners to receive feedback and encouragement from others and cultivate independent abilities in connection with others.

[Post-reading phase]

In this case study, students wrote their understanding of the stories in memos using the keyboard and their handwriting. Currently, the iPad's Speech Recognition function is available for learners who are not good at writing. It is possible for learners to use the voice-to-text function to input what they imagined, understood, and thought about a story into BookRoll memos. This allows for the collection and analysis of data in the LEAF system.

Current reading learning does not end with understanding what was read but requires the ability to develop beyond that and apply information that can be used in real life. These application and practical skills may be enforced through interaction with others. In an inclusive learning environment, learners with and without learning difficulties co-exist. In particular, encouragement from peers may develop learners’ perseverance in the face of challenges and improve their comprehension and learning performance (Gersten et al., 2001 ). For class activities, data-based group formation can be applied in which groups are created to work together to deepen and develop an understanding of what they read. This is possible with the current LEAF, and group formation parameters such as homogeneous, heterogeneous, random, and jigsaw can be adjusted depending on the learning purpose, learner characteristics, and other considerable factors (Liang et al., 2023 ). Further, AI will be able to pair learners who need help with learners who have already completed a task, or create peer help groups based on log data. For example, AI would recommend a human learning companion and/or an AI agent, or called pedagogical agents (Savin-Baden et al., 2019 ), to read together. Peers can be selected from humans or AI in the future, creating an environment that promotes learning and reading together. This may reduce the burden on the teacher in a busy classroom, provide feedback suitable for the individual with the help of AI and the people around it, and manage and orchestrate the class activity efficiently. Depending on the learner’s progress, AI can facilitate a unique inclusive learning experience by potentially involving human intervention and reflection.

AI for facilitating learning reflection and decision-making

Using the LEAF system for AR activities allowed us to capture and visualize participants’ reading processes and detect salient behaviors and insights in learners with special needs. Furthermore, the visualized learning process and artifacts were shared between the resource room teacher and the mother. In the LEAF learning environment, learners can use the dashboard to reflect not only on the results but also on the learning process. Reflection encourages learners’ metacognition by allowing them to reflect on their own thinking, and self-reflection provides an opportunity to evaluate their own cognitive processes (Gersten et al., 2001 ; Silver et al., 2023 ). Generally, learners reflect on their own learning and deepen their understanding, and teachers review their learning and decide what to do next. However, some learners find it difficult to reflect on their own learning. In the AI-driven inclusive education expected in the future, AI may be used to support reflection on reading learning using both text and audio. Using log data from learners’ own learning activities enables more personalized feedback by highlighting interesting and hidden patterns. An AI agent will also play an active role. It will sense “done” or “not done” and provide options for what steps to take while emphasizing what learners can do to increase their self-affirmations. For learners who have difficulty understanding information from graphs and tables, or from texts, audio, and visual images will be automatically selected and added to make it easier for them to understand the information presented on the dashboard to assist in learning comprehension. AI will also automatically explain the data displayed on the dashboard, making it easier to understand not only for learners and teachers but also for parents and other educational supporters. This can improve the efficiency and effectiveness of the decision-making process. For example, learners can decide what to learn next, teachers can choose and plan the next activity, and teachers, school administrators, and parents can decide what kind of support learners will need. AI will further encourage human intervention, making it possible to judge their learning more objectively with the help of stakeholders such as teachers and parents, thus facilitating a unique and comprehensive learning experience.

Data sharing and portability

Data on each student in the SE are necessary to determine the support that should be provided according to the student’s developmental stage. Resource room (and homeroom) teachers are obligated to keep records of students’ learning and progress and to report to the school and parents in accordance with them. Support and data sharing are currently primarily conducted using printouts, which are stored, filed, and shared with parents and schools, along with notes on the teacher’s observations during class. In a data-driven learning environment like LEAF, parents can also use the dashboard to check their child’s growth and objectively consider future support based on logs. One of the potential expectations of a data-driven learning environment is the sharing of learning data widely and throughout life with other stakeholders such as other educational institutions and local governments.

The personal data of learners with special needs are shared and transferred across institutions to ensure that they are adequately supported. Even in the event of a change in the learning environment, such as transferring to a different school, graduating from one institution, or progressing to the next educational stage, past learning and support data can be preserved and transferred upon request. The insights we gained from the teacher interview underscored the significance of the secure and seamless sharing and portability of data. The LEAF system is used by students from elementary schools to universities. It will be possible to safely transfer learning data across multiple learning contexts with the integration of blockchain, such as BOLL (Ocheja et al., 2019 ), and students’ learning logs in BookRoll can be transferred to the next learning context. Further, AI will recommend the relevant schools and/or assist learners in making evaluations and decisions when moving up to higher education or finding employment. However, to enhance safe data sharing and portability, it is necessary to obtain the stakeholders’ understanding of learning using AI technology and enhance the data literacy of teachers and learners as well as that of other stakeholders.

Dissemination and awareness of AI-driven learning

AI has the potential to impact not only students in inclusive education but also teachers and other stakeholders like parents. In today’s learning environment in which education and technology are integrated, teachers are required to possess a wide range of diverse competencies such as technical, pedagogical, and content knowledge (TPACK) to deal with complex learning situations (Mishra & Koehler, 2006 ). According to MEXT ( 2021a ), in order to obtain a teaching license for elementary and junior high school in Japan, all teachers will be required to have practical training regarding special education including nursing care experience, as well as developing data literacy and ICT skills. Past literature has indicated the need for specialized pre-training for learners and teachers (Leshchenko et al., 2020 ; Starks & Reich, 2023 ) and digital literacy and technology (Starks & Reich, 2023 ). The current study further highlighted these needs for teachers and parents. Our findings also implied that learners’ and teachers’ understanding of the potential of new technologies still remains low in Japan, as noted in other countries (DeCoito & Richardson, 2018 ; Hirsto et al., 2022 ; Salas-Pilco et al., 2022 ). We found that not all parents welcome or approve of data-driven learning.

As cited by UNESCO, one of the challenges related to implementing AI in education is transparency and fairness considerations in the collection, use, and dissemination of personal data ( 2019 ). In order to dispel these concerns and gain understanding, it is necessary to disseminate information literacy and provide training not only to learners and teachers but also to other parties involved in supported learning. One of the solutions we suggest includes involving all stakeholders in the learning environment to objectively share a common understanding. This inclusion of stakeholders in the design, development, implementation, and evaluation of systems used for learning could help them understand data- and AI-driven learning, thereby increasing their understanding of its importance. This may also resolve issues such as misunderstandings between stakeholders. To this end, we maintain close contact with local schools, expanding technical and educational support, and continuing to implement supportive and interactive learning.

While some challenges remain, AI-driven learning offers positive impacts for learners, teachers, parents, and all other stakeholders. This pilot study implies that the duties of the resource room teacher were diverse, including, for example, continuously sharing students’ information with other stakeholders like homeroom teachers and parents and providing optimal individualized support to each student. Emerging technologies such as ICT and AI will lead to the efficient management and coordination of class activities, such as improving instruction and creating teaching materials, which will hopefully result in work style reformations. This could include reducing teachers’ workloads and shortening waiting lists of students who are unable to receive support in a resource room due to a lack of human resources and difficulty in coordinating time (MEXT, 2021b ). Furthermore, school administrative support related to special needs education, the creation and sharing of individual education, and various information will become easier, which will directly lead to the improvement of school operations and the enhancement of portability between schools and related organizations. This study highlighted these possibilities through learning with BookRoll and sharing the learning process with teachers and parents on the dashboard. Collaboration with stakeholders expands the learning opportunities for all students in inclusive education.

To date, no study in Japan has investigated the challenges and possibilities of using AI in the context of actual inclusive educational settings from the LA perspective. Therefore, we undertook a case study to explore how an AI-driven approach can materialize the vision of SE as a supportive framework for learners with diverse needs in the context of inclusive education in Japan. In today’s data-enhanced learning environment, it is possible to detect and visualize specific learning behaviors using learning logs obtained from daily learning. By integrating AI technology into the current learning context, we found that individual learners can be provided with more efficient and appropriate learning and reflections on learning. However, while some teachers and parents, such as our participants, look forward to opportunities to objectively reflect on learning and provide further support using AI technology assistance, we realized that obtaining assent and understanding from teachers and parents along with fostering data literacy remains a challenge for future inclusive education utilizing AI.

Our future work includes pursuing the possibilities of an AI-driven inclusive learning environment in which all learners are expected to receive equal learning opportunities and optimal support with the co-progress of stakeholders. This cannot be achieved without a considerable amount of data. In Japan, the GIGA initiative has created an environment for data utilization on a national level. Although it has been pointed out that data utilization has not fully penetrated Japan compared to other countries (MEXT, 2022c ), the country is working to build a large-scale data sphere that supports the use of AI, which has created an environment for the effective use of logs. As the use of educational informatization progresses on a larger scale, the data problems and generalizability concerns found in this study may be resolved. Based on the logs collected from the previous and upcoming implementations, we will derive an AI algorithm that will realize and aim to create an AI-driven inclusive learning environment that can provide individually optimal learning support to each learner in cooperation with stakeholders. From there, we will pursue evaluating the impact of AI and understanding the actual situations for inclusive education.

Availability of data and materials

The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Abbreviations

Attention deficit hyperactivity disorder

Artificial intelligence

• Active reading

Blockchain of learning logs

Developmental disorders (or disabilities)

Global and Innovation Gateway for ALL

Information and Communication Technology

• Learning analytics

Learning difficulties

Learning & Evidence Analytics Framework

Learning management system

Learning Record Store

Ministry of Education, Culture, Sports, Science and Technology

Organisation for Economic Co-operation and Development

Programme for International Student Assessment

• Special education

Special needs education

Survey, Question, Read, Recite, and Review

Survey, Question, Read, Record, Recite, and Review

Technological, Pedagogical, and Content Knowledge

Universal Design for Learning

Bodily, R., & Verbert, K. (2017). Trends and issues in student-facing learning analytics reporting systems research. In Proceedings of the seventh international learning analytics & knowledge conference (pp. 309–318).

Bryant, D. P., Bryant, B. R., & Smith, D. D. (2019). Teaching students with special needs in inclusive classrooms . Sage Publications.

Google Scholar  

Buzzi, M. C., Buzzi, M., Perrone, E., Rapisarda, B., & Senette, C. (2016). Learning games for the cognitively impaired people. In Proceedings of the 13th international web for all conference (pp. 1–4).

Cano, A., & Leonard, J. D. (2019). Interpretable multiview early warning system adapted to underrepresented student populations. IEEE Transactions on Learning Technologies, 12 (2), 198–211.

Article   Google Scholar  

DeCoito, I., & Richardson, T. (2018). Teachers and technology: Present practice and future directions. Contemporary Issues in Technology and Teacher Education, 18 (2), 362–378.

Dillenbourg, P. (2013). Design for classroom orchestration. Computers & Education, 69 , 485–492.

Gersten, R., Fuchs, L. S., Williams, J. P., & Baker, S. (2001). Teaching reading comprehension strategies to students with learning disabilities: A review of research. Review of Educational Research, 71 (2), 279–320.

Hirsto, L., Valtonen, T., Saqr, M., Hallberg, S., Sointu, E., Kankaanpää, J., & Väisänen, S. (2022). Pupils’ experiences of utilizing learning analytics to support self-regulated learning in two phenomenon-based study modules. In Society for information technology & teacher education international conference (pp. 1682–1688). Association for the Advancement of Computing in Education (AACE).

Idol-Maestas, L. (1985). Getting ready to read: Guided probing for poor comprehenders. Learning Disability Quarterly, 8 (4), 243–254.

Kazimzade, G., Patzer, Y., & Pinkwart, N. (2019). Artificial intelligence in education meets inclusive educational technology—The technical state-of-the-art and possible directions. In Artificial intelligence and inclusive education: Speculative futures and emerging practices (pp. 61–73).

Kinoshita, T., Imu, Y., & Ishida, S. (2023). [A research trend on the use of ICT in special needs education: Focusing on intellectual and developmental disabilities] Tokubetsushienkyoiku niokeru ICT no rikatsuyo ni kansuru kenkyudoko (in Japanese). Bulletin of the Faculty of Education Chiba University, 71 , 107–115.

Kumagai, H., & Nagai, N. (2022). [Characteristics of information literacy of children attending resource room—Analysis through development and application of an information literacy checklist] Tsukyusidokyoshitu wo riyosuru jido niokeru jyohokatsuyonouryoku no tokucho: Jyohokatsuyonoryoku checklist no sakusei to chosa wo toshite (in Japanese). Bulletin of Miyagi University of Education Graduate School of Teacher Education, 3 , 147–156.

Leshchenko, M., Tymchuk, L., & Tokaruk, L. (2020). Digital narratives in training inclusive education professionals in Ukraine. In J. Głodkowska (Ed.), Inclusive education: Unity in diversity (pp. 254–270). Akademii Pedagogiki Specjalne.

Liang, C., Toyokawa, Y., Majumdar, R., Horikoshi, I., & Ogata, H. (2023). Group formation based on reading annotation data: system innovation and classroom practice . Journal of Computers in Education , 1–29.

Margetts, H., & Dorobantu, C. (2019). Rethink government with AI. Nature, 568 (7751), 163–165.

Mejia, C., Florian, B., Vatrapu, R., Bull, S., Gomez, S., & Fabregat, R. (2016). A novel web-based approach for visualization and inspection of reading difficulties on university students. IEEE Transactions on Learning Technologies, 10 (1), 53–67.

Ministry of Education, Culture, Sports, Science and Technology. (2012). Promotion of special needs education for building an inclusive education system toward the formation of a cohesive society (report) overview. Retrieved September 21, 2023, from https://www.mext.go.jp/b_menu/shingi/chukyo/chukyo3/044/attach/1321668.htm

Ministry of Education, Culture, Sports, Science and Technology. (2020). A guide for teachers in charge of resource room instruction for the first time. MEXT Elementary and Secondary Education Bureau Special Needs Education Division. Retrieved November 21, 2023, from https://www.mext.go.jp/tsukyu-guide/common/pdf/passing_guide_02.pdf

Ministry of Education, Culture, Sports, Science and Technology. (2021a). Report from the expert meeting on the new era of special needs education. Retrieved November 21, 2023, from https://www.mext.go.jp/content/20210208-mxt_tokubetu02-000012615_2.pdf

Ministry of Education, Culture, Sports, Science and Technology. (2021b). Aiming to build “Japanese-style school education in the Reiwa era”—Realizing optimal individual learning and collaborative learning that brings out the potential of all children—(Report). Retrieved November 21, 2023, from https://www.mext.go.jp/content/20210126-mxt_syoto02-000012321_2-4.pdf

Ministry of Education, Culture, Sports, Science and Technology. (2022a). Regarding the survey results (2020) regarding children enrolled in regular classes who require special educational support. Retrieved November 21, 2023, from https://www.mext.go.jp/b_menu/houdou/2022/1421569_00005.htm

Ministry of Education, Culture, Sports, Science and Technology. (2022b). Results of a survey on the implementation status of instruction for resource room (overview). Retrieved November 21, 2023, from https://www.mext.go.jp/content/20220905-mxt_tokubetu01-000023938-10.pdf

Ministry of Education, Culture, Sports, Science and Technology. (2022c). Overview of AI strategy 2022: April 2020 cabinet office science, technology and innovation promotion secretariat. Retrieved November 21, 2023, from https://www8.cao.go.jp/cstp/ai/aistrategy2022_gaiyo.pdf

Mishra, P., & Koehler, M. J. (2006). Technological pedagogical content knowledge: A framework for teacher knowledge. Teachers College Record, 108 (6), 1017–1054.

National Institute for Educational Policy Research. (2022). OECD student assessment (PISA). Retrieved September 21, 2023, from. https://www.nier.go.jp/kokusai/pisa/

Ocheja, P., Flanagan, B., Ueda, H., & Ogata, H. (2019). Managing lifelong learning records through blockchain. Research and Practice in Technology Enhanced Learning, 14 (1), 1–19.

Ogata, H., Majumdar, R., Akçapinar, G., Hasnine, M. N., & Flanagan, B. (2018). Beyond learning analytics: Framework for technology-enhanced evidence-based education and learning. In 26th international conference on computers in education workshop proceedings (pp. 493–496). Asia-Pacific Society for Computers in Education (APSCE).

Peterson, C. L., Caverly, D. C., Nicholson, S. A., O’Neal, S., & Cusenbary, S. (2001). Building reading proficiency at the secondary level: A guide to resources. Introduction.

Randi, J., Newman, T., & Grigorenko, E. L. (2010). Teaching children with autism to read for meaning: Challenges and possibilities. Journal of Autism and Developmental Disorders, 40 , 890–902.

Rose, D. H., & Meyer, A. (2002). Teaching every student in the digital age: Universal design for learning . Association for Supervision and Curriculum Development (Product no. 101042: $22.95 ASCD members; $26.95 nonmembers).

Rose, D. R. (2019). Students with learning disabilities and their perspectives regarding reading comprehension instruction: A qualitative inquiry. Journal of Ethnographic & Qualitative Research, 14 (2), 137–152.

Ryan, J. (2007). Learning disabilities in Australian universities: Hidden, ignored, and unwelcome. Journal of Learning Disabilities, 40 (5), 436–442.

Salas-Pilco, S. Z., Xiao, K., & Oshima, J. (2022). Artificial intelligence and new technologies in inclusive education for minority students: A systematic review. Sustainability, 14 (20), 13572.

Savin-Baden, M., Bhakta, R., Mason-Robbie, V., & Burden, D. (2019). An evaluation of the effectiveness of using pedagogical agents for teaching in inclusive ways. In Artificial Intelligence and Inclusive Education. Speculative Futures and Emerging Practices (pp. 117–134).

Siemens, G., & Baker, R. S. D. (2012). Learning analytics and educational data mining : towards communication and collaboration. In Proceedings of the 2nd international conference on learning analytics and knowledge (pp. 252–254).

Silver, N., Kaplan, M., LaVaque-Manty, D., & Meizlish, D. (Eds.). (2023). Using reflection and metacognition to improve student learning: Across the disciplines, across the academy . Taylor & Francis.

Slowík, J., Gažáková, E., Holeček, V., & Zachová, M. (2021). Comprehensive support for pupils at risk of school failure in inclusive education: Theory and school practice in the Czech Republic. International Journal of Inclusive Education, 27 , 1–17.

Starks, A. C., & Reich, S. M. (2023). "What about special ed?“: Barriers and enablers for teaching with technology in special education. Computers & Education . https://doi.org/10.1016/j.compedu.2022.104665

Toyokawa, Y., Majumdar, R., Kondo, T., Horikoshi, I., & Ogata, H. (2023). Active reading dashboard in a learning analytics enhanced language-learning environment: effects on learning behavior and performance. Journal of Computers in Education , 1–28.

Toyokawa, Y., Majumdar, R., & Ogata, H. (2022). Learning analytics enhanced E-book reader in a Japanese special needs class. In 2022 International conference on advanced learning technologies (ICALT) (pp. 274–278). IEEE.

UNESCO. (2009). Policy guidelines on inclusion in education. Retrieved September 21, 2023, from https://unesdoc.unesco.org/ark:/48223/pf0000177849

UNESCO. (2019). Artificial intelligence in education: challenges and opportunities for sustainable development. Retrieved September 21, 2023, from https://unesdoc.unesco.org/ark:/48223/pf0000366994

Yin, R. K., & Moore, G. B. (1987). The use of advanced technologies in special education: Prospects from robotics, artificial intelligence, and computer simulation. Journal of Learning Disabilities, 20 (1), 60–63.

Zainal, Z. (2007). Case study as a research method. Jurnal Kemanusiaan , 5 (1).

Download references

Acknowledgements

The authors would like to thank the resource room teacher and the parents for their contributions to the study.

This work is partially funded by NEDO JPNP20006 and JPNP18013, JSPS KAKENHI (A) JP23H00505 and (B) JP22H03902 and, National Institute for Educational Policy Research: Educational Data Analysis and Research Promotion Project FY2023-2025.

Author information

Authors and affiliations.

Graduate School of Informatics, Kyoto University, Yoshidahonmachi, Sakyo Ward, Kyoto, 606-8501, Japan

Yuko Toyokawa

Academic Center for Computing and Media Studies, Kyoto University, Yoshidahonmachi, Sakyo Ward, Kyoto, 606-8501, Japan

Izumi Horikoshi, Rwitajit Majumdar & Hiroaki Ogata

Research and Educational Institute for Semiconductors and Informatics, Kumamoto University, 2-39-1 Kurokami, Chuo-Ku, Kumamoto City, Kumamoto, 860-0862, Japan

Rwitajit Majumdar

You can also search for this author in PubMed   Google Scholar

Contributions

YT conceived the idea of the study, conceptualized it, performed the data analysis, and drafted the original manuscript. RM contributed to the discussion on the contribution of AI to special needs education, and IH contributed to conducting the analysis. HO initiated the framework of the overall argument and supervised the conduct of this study. RM and HO acquired funding for the research. The authors read and approved the final manuscript.

Corresponding author

Correspondence to Yuko Toyokawa .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Toyokawa, Y., Horikoshi, I., Majumdar, R. et al. Challenges and opportunities of AI in inclusive education: a case study of data-enhanced active reading in Japan. Smart Learn. Environ. 10 , 67 (2023). https://doi.org/10.1186/s40561-023-00286-2

Download citation

Received : 29 September 2023

Accepted : 08 December 2023

Published : 18 December 2023

DOI : https://doi.org/10.1186/s40561-023-00286-2

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Inclusive education

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List
• Pan Afr Med J

Real-world application, challenges and implication of artificial intelligence in healthcare: an essay

Shiv kumar mudgal.

1 College of Nursing, All India Institute of Medical Sciences, Deoghar, Jharkhand, India,

Rajat Agarwal

2 Department of Cardiothoracic Surgery, All India Institute of Medical Sciences, Deoghar, Jharkhand, India,

Jitender Chaturvedi

3 Department of Neurosurgery, All India Institute of Medical Sciences, Rishikesh, Uttarakhand, India

Nishit Ranjan

This essay examines the state of Artificial Intelligence (AI) based technology applications in healthcare and the impact they have on the industry. This study comprised a detailed review of the literature and analyzed real-world examples of AI applications in healthcare. The findings show that major hospitals use AI-based technology to enhance knowledge and skills of their healthcare professionals for patient diagnosis and treatment. AI systems have also been shown to improve the efficiency and management of hospitals´ nursing and managerial functions. Healthcare providers are positively accepting AI in multiple arenas. However, its applications offer both the utopian (new opportunities) as well as the dystopian (challenges). Unlike pessimists, AI should not be seen a potential source of “Digital Dictatorship” in future of 22 nd century. To provide a balanced view on the potential and challenges of AI in healthcare, we discuss these details. It is evident that AI and related technologies are rapidly evolving and will allow care providers to create new value for patients and improve their operational efficiency. Effective AI applications will require planning and strategies that transform both the care service and the operations in order to reap the benefits.

Introduction

Artificial Intelligence (AI), a technology prevalent for almost 60-year has made it possible to create applications that have a profound effect on our life today. It seeks to reproduce and modify human intelligence leading to development of intelligent machines [ 1 ]. Some researchers believe that AI can think and act rationally. Others disagree that AI is capable of acting and thinking like humans. Irrespective of what anyone believes, it appears for sure that in the year 2100, the health industry is expected to survive on AI-Human cooperation, not competition. Artificial intelligence, a broad-based tool, allow humans to rethink the way they integrate information, analyze data and use the insights to improve their decision-making. It is already transforming all walks of life [ 2 ].

AI is not something futuristic, but a technology that is already in use and integrated into many sectors. Examples include public healthcare and education, transport, telecommunications, data security management, finance, research, policymaking and the legal and judiciary system. AI technologies are now being increasingly applied to healthcare [ 3 ]. A combination of unstoppable forces drives healthcare demand. These include changing patient expectations, increasing population age, lifestyle shifts, and the never-ending circle of innovation. The Healthcare system must undergo significant structural and transformational changes to ensure its sustainability. AI has potential to transform healthcare and address some of these challenges [ 4 , 5 ].

AI has been welcomed by healthcare systems around the world, which struggle to fulfil the “quadruple objective” of improving the health and well-being of their patients, healthcare access, cost-effectiveness [ 6 ] and improving the lives of healthcare workers [ 7 ]. It is essential for healthcare providers to be well versed in the potential applications of AI technologies in different aspects of healthcare which may embark digital revolution in this sector [ 8 ]. This article will discuss numerous applications and issues of AI technology in the healthcare industry in the present times. The article also serves necessary recommendations which will help healthcare managers with strategic planning and execution of AI in healthcare.

Operational terms

What is ai.

UNESCO defines AI systems as “technological systems that can process information in a manner that resembles intelligent behavior” [ 9 ]. A simplified definition of AI for healthcare is the ability to use computer programs to perform tasks or reasoning in multiple areas of healthcare, including diagnosis and treatment. This is similar to the intelligence that we associate with intelligence in humans [ 10 ]. AI in healthcare also refers to the use of machine-learning algorithms or software to replicate human cognition in the analysis and presentation of complex medical and healthcare data [ 11 ].

Types of AI

The main categories of AI are based on the capabilities and functions of AI. The types of AI are explained in the diagram below ( Figure 1 ).

types of artificial intelligence

Based on capabilities of AI-based systems

i) Weak AI or Narrow AI: an AI that can perform a specific or a limited set of tasks without any thinking abilities. The Weak AI category covers almost all AI-based systems that have been developed to date such as Siri, Alexa Self-driving car, Alpha-Go, Sophia the humanoid and speech recognition agent [ 12 , 13 ]. ii) General AI or Strong AI - It can perform any intellectual task as efficiently as a human. Although there are no examples of Strong AI to date, we believe that it will soon become possible to build machines as smart and intelligent as humans [ 12 , 13 ]. iii) Super AI: this level of intelligence of systems is at which machines can surpass human intelligence and have cognitive properties. It is currently a fictional scenario [ 12 , 13 ].

Based on the functionality of AI-based systems

i) Reactive machine AI: this category of AI includes machines that work solely from current data and take only current circumstances into consideration. It cannot draw inferences to predict their future actions. They can do a narrower variety of pre-defined functions. Examples include Google's Alpha Go and IBM's Deep Blue systems [ 12 , 13 ]. ii) Limited Memory AI: this has limited memory. It can make better decisions by looking at past data, can store past experiences in a short-term or temporary manner and then use that information to predict future actions. Example includes self-driving vehicles [ 12 , 13 ]. iii) The Theory of Mind AI: this AI should understand people, emotions, and beliefs of human beings, and can interact with them socially as humans. This type of AI machine is not yet developed [ 14 , 15 ]. iv) Self-aware AI: self-awareness AI is the future. This technology will build super-intelligent machines that will possess their own consciousness, feelings, and self-awareness. These machines will be more intelligent than the human brain. It´s only a fictional scenario at present [ 14 , 15 ].

The state of AI technology

Artificial intelligence does not refer to one technology. Many AI technologies can be applied immediately to healthcare. Some great AI technologies in healthcare are mentioned below:

Machine learning (ML) : machine learning is the dominant approach to AI. It uses a predictive model for making predictions from predefined data. Machine learning is being used in many AI technologies, such as natural language processing (NLP) and voice technology [ 8 ]. Supervised learning, reinforcement learning (RL), deep learning and multi-instances learning are some of the most popular ML algorithms [ 14 ].

Supervised learning : this approach uses a set of data and known, defined outcomes as an outcome. Then, patterns are identified that correspond with the input to make predictions. The algorithm must know what conclusions it should draw from the given data set. Healthcare has witnessed a lot of supervised learning. This allows for data-driven clinical decisions to be made, e.g., use of imaging to diagnose tumours and determine their severity, and predictive analytics within continuous outputs e.g., use of EHR to predict the recurrences, prognosis and mortality of a particular disease [ 14 , 15 ].

Unsupervised learning : this approach can find the data structure and forecast based only on that input. It is better suited for uncertain outcomes or when data labelling is expensive. Unsupervised learning can be used in healthcare to predict individual disease risks and design personalized treatments that are based on genetic biomarkers and genomic variations [ 8 , 15 ].

Semi-supervised learning : unsupervised learning is able to learn by itself, without any human interventions for the outcome. Unsupervised learning, even without human instruction, can be more susceptible to errors as it may use minor features of the data for predictions. In practice, semi-supervised learning is often used. It uses a combination of large untagged and small tagged data for training [ 15 ].

Reinforcement learning : it is an autonomous algorithm that allows the user to act and interact according to the environment. It is one of the best learning models and very effective for tasks with clearly defined protocols. It relies on its own experience using feedback from mistakes and rewards to lead training. It does not require data or labelling. It is useful in healthcare such as optimizing treatment plans and robotic-assisted surgical procedures [ 15 , 16 ].

Deep learning : deep learning uses a backpropagation algorithm that operates on multiple levels of abstraction to uncover the complex structure of large datasets. This algorithm is designed for the solution of difficult practical world issues. Some examples are: computer vision, Go game, speech recognition, NLP, genomics and drug discovery [ 8 , 16 ].

Natural language processing : this employs a computational approach to automatically interpret and represent human language, mainly in text form. These include machine translation, speech recognition, speech classification, question answering and sentiment analysis. Natural language processing tools can extract vital information about patients from large textual data like doctor´s prescriptions, daily patient notes, discharge summaries and various radiological / laboratory reports. This can help healthcare providers in speedy management of patients thus optimizing the health care delivery [ 8 , 16 ].

Real-world applications : the meaningful and practical application of AI, provides healthcare providers with opportunities and confidence to boost their skills to new challenges in healthcare.

Real-World AI applications in healthcare

Some of the important practical applications in healthcare are mentioned in the following sections:

Precision medicine : one of the most important applications of AI in healthcare is precision medicine. Precision medicine aims to optimize the path for diagnosis, therapeutic intervention and prognosis. It uses large multidimensional biological data sets that capture individual variability in genes and other contributing factors like age, gender, and race, as well as medical treatment options such as immune profile, metabolism and vulnerability to the environment. This allows clinicians to tailor early interventions, whether preventative or treatment-oriented, to each patient. There are many precision medicine initiatives [ 17 ]. These can be divided into three categories: digital health apps and complex algorithms, as well as genomic-based tests [ 17 ]. A deep learning algorithm was developed in collaboration with Scripps Research Institute (CA, USA) and Intel. With a precision of 85%, it could identify 23 patients at high risk for cardiovascular disease. This cognitive assistant is equipped with clinical knowledge and reasoning [ 18 ].

Improved disease treatment : AI technologies are increasingly adding to the support of healthcare workers in various aspects of patient´s management. For instance, Onduo offers virtual coaching on mobile apps to control diabetes. It employs AI technology to detect food, and monitor glucose levels as well as physical activities, in order to make recommendations. DayTwo provides another solution for diabetes management. It provides an individualized meal suggestion, based on the user's gut microflora for adequate blood sugar levels. The recommended diet is chosen from its large index of over 100,000 foods [ 19 ]. ResApp Health, another example of AI used in chronic disease management, analyzes subjects' breathing by using their phone microphone. The AI algorithm then evaluates various respiratory conditions like chronic obstructive lung disease, pneumonia accurately [ 20 ].

Improved diagnostic error reduction and decision support: AI will be used to aid in diagnosing patients with certain diseases and reduce human errors. AI was used by the Mayo Clinic to screen for cervical cancer in order to detect pre-cancerous changes. To identify precancerous signs, the AI-based algorithm uses over 60,000 images of cervical cancer from the National Cancer Institute. The accuracy rate of the algorithm was 91% as compared to 69% by skilled human expert [ 21 ]. The focus of IBM's Watson for Oncology has been a focus of media, especially in oncology management. Watson uses combination machine learning and NLP capabilities [ 22 ]. Freenome, which uses molecular biology and machine learning to detect early-stage cancers, is another example. The model can be trained to identify which biomarker patterns indicate the stage, type and best treatment options for particular cancer. AI can be used to detect disease-associated patterns by decoding hidden patterns. Google health uses AI for breast cancer screening. It demonstrated that its AI system can outperform human experts in breast-cancer prediction [ 23 ]. A deep learning-based AI developed by Massachusetts Institute of Technology (MIT), can forecast the possibility of development of breast cancer up to five years ahead [ 23 ].

London´s Moorefield´s Eye Hospital, has declared an AI solution for identifying ocular disease. The AI-based algorithm used data from greater than 15,000 British patients to detect ocular diseases by optical coherence. The decision of referral made by the AI-based algorithm was 94% accurate [ 24 ]. Google's research team developed a deep learning algorithm that can interpret retinal images to identify signs of diabetic retinopathy. This could potentially help doctors screen more patients in areas with fewer resources [ 18 ]. There are between 6000 and 8000 rare diseases that affect approximately 400 million people around the globe. A rare disease diagnosis can take up to five years and is often time-consuming having a great impact on the finances of the patient and the system. 3Billion created an algorithm in 2019 to diagnose rare DNA-based conditions which can test for up to 7000 diseases simultaneously in suspected cases [ 25 ].

Patient data analytics : AI allows hospitals for clinical data analysis which can provide in-depth of patient´s health. It can also be used to predict prognosis, help in clinical audits, track patient prescription and refills, predict the advantages of specific drugs and identify patients´ at risk for substance abuse [ 26 ]. For example, the Paris public university hospital uses the Intel analytics platform for predicting the number of patients visiting the emergency department [ 27 - 29 ]. The potential volume of data is huge. According to estimates, personal lifestyle-based data amount to approximately 1100 terabytes in a lifetime. Genetics and medical data account for 6.4 terabytes. Omics technology, GWAS and EWAS, smartphone-based digital phenotyping, sensors and EHRs, and wearable devices can accurately monitor the lifestyle of a person along with climate and topographical data. This made it possible to implement strategies for the prevention and management of metabolic lifestyle disorders. This is why structured data collection and analysis are necessary for large, multidimensional studies which can be employed by integration ML/AI in healthcare system [ 29 , 30 ].

Medical robotics : medical robots have many uses. They can be used to assist in surgery, in rehabilitation for stroke patients (rehabilitation robotics), care for elderly persons (assist-living robotic companion) social interaction (humanoid robot) and so on. AI-assisted surgeon robots have found their way into operation theatres. They can perform surgeries without fatigue and very useful at places where human hands cannot operate due to space constraints [ 27 , 31 ]. The Da Vinci is a surgical robot that allows professionals to perform complex procedures with greater flexibility and control than traditional approaches. The Da Vinci is a surgical robot that can assist surgeons by translating their hand movements at the console and creating clear, magnified, 3D high-resolution images of the surgical site [ 32 ].

Real-time prioritization and triage : triage machine learning has been shown to be an efficient tool. John Hopkins University researchers found that ML-based e-triage improves patient risk assessment and categorization. Enlitic is patient triaging software that prioritizes cases according to their clinical data and directs them to suitable medical personnel [ 27 ]. Babylon health provides applicable health and triage information depending upon symptoms of the patient [ 33 ].

Personalized care or virtual assistance : the treatment plans based on patient data reduce cost and increase the effectiveness of care. Human-Machine Interfaces (HMIs) analyze and recognizes facial motions and helps person with disabilities to drive robotic vehicles and wheelchairs [ 34 ]. RUDO, an “ambient intelligent system”, can be used to help blind people live with sighted people and work in trained fields like computer science. Blind people can access the various functions of the virtual assistant through a single interface [ 35 ]. An AI-based smart assistant can advise pregnant mothers about various important antenatal matters. AI applications can help the elderly with routine medications and can predict and prevent falls. This can be of major help in patients with gait disorders like Parkinson´s disease [ 28 ]. Chatbots allow patients to self-diagnose or help physicians in making a diagnosis. They can help patients share their health information in a proactive way. This allows medical professionals to improve quality care with cost-effectiveness. It also helps to increase patient satisfaction [ 27 ]. GYANT is a chatbot for healthcare that helps patients understand their symptoms. Doctors then receive the data and can diagnose and prescribe medicines in real time. Woebot is another chatbot that focuses on mental health. It calls itself “the next generation of mental health” and it certainly seems that way. The chatbot uses Cognitive Behavioral Therapy or CBT, to listen and offer advice to anyone who seeks it out [ 36 ]. AI apps that monitor and assist patients´ compliance to prescribed medication and treatment have been proven to be effective. Sentrian uses AI to analyze the data collected from patients' sensors at home. The goal is to identify signs and conditions that could lead to deterioration early so that intervention can be taken to prevent hospital admissions [ 37 ].

Virtual assistants for nursing : AI virtual assistants are great in nursing because they can keep healthcare providers and patients connected all the time and thus decreases pressure on the already overburdened medical staff. Alexa robots are virtual nursing assistants employed by Cedars-Sinai Hospital in Los Angeles, California help nursing staff in their daily chores. [ 22 ]. Sensely, a virtual nurse, use Natural Language Processing, Machine Learning and wireless integration to medical devices, such as blood pressure monitors, to provide assistance to patients. Sensely can help you with self-care and clinical advice. It also helps you to schedule an appointment [ 38 ].

Administrative workflow assistance : one of the AI applications in healthcare is the automation of administrative workflow. AI systems are able to perform operations like the transcription of medical records, medical billing services, bed allotment, and insurance claim verifications apart from numerous other hospital administrative activities faster and much more accurately than humans.[ 22 , 38 ] faster and better than individuals. The AI robot Paul accompanies the medical personnel in their daily patient rounds, help in the analysis of patient medical records and can provide any information regarding patient including daily investigations in a fraction of a second. Maria, the robot's guide, provides directions to patients in the hospital lobby to their doctor's offices or specific medical departments within the hospital and schedules appointments by touching the robot with their medical ID card [ 39 ]. The official statement made by Johns Hopkins University Hospital regarding AI technology stated: “Emergency room patients are assigned beds 30% faster, transfer delays from operating rooms are reduced by 70%, ambulances can pick up patients from other hospitals 63 minutes earlier, and the ability to take patients with complex medical conditions from regional and national hospitals has improved to 60%” [ 40 ]. Microsoft´s AI digital Assistant Cortana employed advanced analytics and predictive technology to identify potential patients at-risk in ICU treatment and able to monitor “100 beds in 6 ICUs”. [ 40 ].

Improved operational efficiency and cost effectiveness : AI-based medical systems can perform numerous tasks involved in healthcare services in a simplified and cost-effective manner. Some of the tasks can be even done without human support. An AI-integrated pill-cam can substitute conventional upper endoscopy. Escalante et al . developed an AI-based method for diagnosis of acute leukemia by examining bone marrow structure characteristics non-invasively [ 41 ].

Improving biomedical research : AI acts as an “eDoctor” to diagnose, manage, and prognosis diseases. AI can be a great tool for the indexing of medical literature. It can be used to formulate a research question, search available literature within seconds and test scientific hypotheses. This can save a lot of time and allow the researchers to perform good studies with relevant conclusions in shortest possible time [ 28 , 42 ].

Drug discovery : deep learning has many promising applications in drug discovery. These include advanced image analysis, prediction of molecular structure, function and automated generation of unique chemical entities, de novo drug design, prediction of drug activity, prediction of drug-receptor interactions and prediction regarding drug reaction [ 43 ]. NuMedii, a Biopharma firm, developed an AIDD technology (Artificial Intelligence for Drug Discovery), that can identify rapid connections between drugs, diseases, and systems, if any [ 27 ]. Researchers created Eve, an AI “robot scientist” that is meant to speed up the process of drug discovery in a more economical way [ 44 ].

Potential challenges of the application of AI to the healthcare industry

Some of the most significant challenges in the widespread use of AI include:

Data privacy and cyber security : privacy issues can arise when confidential patient data is collected and shared by AI-based systems/technologies on large datasets. Thus, it is important that AI technology must follow, medical ethics, and laws and should be governed by some laws [ 22 ]. The highly sensitive confidential data of patients can be accessed and manipulated by miscreants who may be detrimental to the patient´s social life. Also, there may be high chances of misdiagnosis because of wrong faked data by AI systems. One study showed that benign moles could be misdiagnosed as malignant simply by adding antagonistic noises or just rotation [ 45 ].

Reliability and safety : any error made by AI system, if not rectified early can lead to wrong results of the assigned tasks which may have serious consequences. For example, an AI app used for predicting the likelihood of patients developing complications after pneumonia wrongly advised doctors to send asthmatic patients home [ 46 ].

Accountability of technology use : if AI-based technology used by medical staff leads to the death of the patient, “who would be responsible for the outcome?” This will create multiple unanswered questions on many technical, managerial and ethical issues [ 22 ].

Potential loss of support system and autonomy : AI health apps may empower individuals to manage their own symptoms and take care of their own needs as and when required. This can have a potential impact on the employment of healthcare workers. This can also lead to less dependency on family members and can lead to isolation and behavioural issues [ 47 ]. AI agents could affect individual autonomy negatively by narrowing the treatment options and thus restricting patients to make informed consent about the procedure [ 45 ].

Challenges in generalization to new populations : AI systems are still far from being able to provide reliable generalizability or clinical application for most types of medical data [ 45 ].

Technological challenges : AI models are usually developed by non-medical professionals and thus end users (healthcare providers and patients) have no control in the derivation of the results. This lack of transparency is one of the major challenges in front of government policymakers. Another challenge is AI technology's limitations as they are designed by humans and any minute error in designing AI system can lead to wrong results. In addition, AI systems are not able to handle unstructured information such as medical imaging, which makes up a significant chunk of medical data in healthcare. Lastly, there is no standardization of data which is to be fed into databases and this can lead to different results in different locations [ 47 ].

Organizational and managerial challenges : there are various challenges in developing AI like exchange and possession of data along with the potential danger of losing skilled healthcare providers and ground-level workers [ 41 ].

Malicious use : although AI can be used to benefit humanity, it is also susceptible to being used maliciously. AI can be used to covertly monitor and analyze motor behaviours that can reveal the identity and secret information of the involved person [ 43 ].

Conclusion: in today´s digital age, innovation is essential. AI and related technologies can be very useful adjuncts to healthcare leaders in various aspects of healthcare management. They should not be viewed as a substitute for medical personnel but as a growing necessity that industries must embrace in order to have a competitive advantage. Artificial Intelligence and Human Stupidity run side by side to improve life of none other than stupid humans. AI over shine its master in two important aspects: connectivity and updatability. Because of its transformative nature in healthcare, the healthcare industry is particularly subjected to the potential of AI applications. AI applications have the potential to change not only the treatment and diagnosis processes but also the lifestyles of patients. In this study, we examined the impact AI technology on healthcare, as well as the types of new challenges and opportunities it has provided. We also recommend the establishment of a legal and ethical structure for AI, and drawing a social consensus between all stakeholders.

Cite this article: Shiv Kumar Mudgal et al. Real-world application, challenges and implication of artificial intelligence in healthcare: an essay. Pan African Medical Journal. 2022;43(3). 10.11604/pamj.2022.43.3.33384

Competing interests

The authors declare no competing interests.

Authors' contributions

Shiv Kumar Mudgal, Rajat Agarwal, Jitender Chaturvedi, Rakhi Gaur and Nishit Ranjan participated in the conceptualization of the study, sourced the materials, and drafted the manuscript. All authors read and approved the final version of the manuscript.

Home › Study Tips › Artificial Intelligence case study

Artificial Intelligence case study

• Published November 2, 2022

Table of Contents

A case study of artificial intelligence.

In a world where more people have a keen interest in artificial intelligence, we want to know what AI looks like in the real world – its threats, challenges, opportunities and solutions to modern-day human problems.

Can artificial intelligence really help humans thrive? And if so, what might be the common downfalls of depending on AI in certain industries?

In this article, we’ll take look at one artificial intelligence case study to begin to form insight into this compelling question. 

What is artificial intelligence? Case study artificial intelligence

Artificial intelligence, or AI, is the theory and development of computer systems that are able to perform tasks normally required by human intelligence. Some examples of these tasks include speech recognition, visual perception, decision-making, and translation between languages. 

In other words, every time you say “Hey Alexa..” you’re using AI. But in what other areas can we find AI in our lives? Take a look at these common examples you’re bound to already be familiar with:

• Netflix uses AI to determine streaming suggestions based on your viewing history
• Facebook uses all the data you input on the platform, from the videos you watch to what you say in your status update, to determine which advertisements you might be interested in
• Universities use essay submission software to determine if work has been plagiarized
• Google Maps utilizes ongoing satellite imagery to determine the best route for you to take on a given journey

From the above examples, you can see how artificial intelligence is now less a figment of a mere Sci-Fi novel and something we commonly interact with in our everyday lives, often without even thinking about it.

But scientists, scholars and innovators are keen to learn more about AI and what it can do on a more complex level.

Human Brain Chips, Elon Musk’s NeuroLink – An AI case study

Scholars have long been interested in how the brain works. Neuroscientists in particular have a vested interest in understanding the human brain, what makes it tick, and the causes and solutions to common conditions that limit a person’s uses of their brain and bodily functions.

The last two decades has seen significant increased interest in the realm of neurotechnology. In 2008, a monkey with an implant was successfully able to control a robotic arm to feed itself through activity in the brain, and as a result, in 2012, the first human brain-controlled robotic arm became a success. In 2017, a paralyzed human was able to control a cursor mentally to type out words and sentences on a computer and in 2018, that same person was able to use a tablet functionally to browse the web, send emails and play games. 

In 2019, Neurolink, a private company founded by famous billionaire and CEO of Tesla, Elon Musk, introduced further advancements in AI brain technology with a pig named Gertrude. 

Gertrude had a wireless device implanted in her brain that was able to monitor a thousand neurons at a time, a significant advancement in neuroscience technology that could potentially become another tool for understanding the brain, as well as lead to other technological advancements. Prior to this device, only 300 neurons could be transmitted at a time, therefore this piece of tech was pretty ground-breaking.

From the pig experiment, it became clear to the world that Neurolink was seriously invested in this area of neurotechnology and had the tools and vision to potentially advance AI beyond what it had been capable of up to that point in time.

“The initial goal of our technology is to help people with paralysis regain independence through the control of computers and mobile devices.” Neurolink states on their website. “Our devices are therefore currently being designed to one day give people the ability to communicate more easily via text or speech synthesis, to follow their curiosity on the web, or to express their creativity through photography, art, or writing apps.” Neurolink. 

In April 2021, another marvel was presented to the world by the company in the form of a real live macaque monkey that demonstrated its ability to play a video game called Mind-Pong using only brain power thanks to their new N1 device and pager. The monkey was able to play the game successfully with only its mind.

This communication from the brain to the screen was made possible through a small device and pager implanted into the monkey’s brain that essentially translated the primate’s synaptic input to initiate an action. In other words, the device was able to tell the technology what to do based on the messages received from the monkey’s brain activity. Sounds like science fiction right?

Neuroscientist Dr Paul Nuyujukian stated that “there was definitely a lot of clever engineering that went into that. To build a device, that can transmit 2,048 electrodes worth of spiking information.. Over a radio, wirelessly…When you have that many channels the performance that you should be able to get should be eclipse what we’ve been able to do in the academic field.” 

On the flip side of the advancement, however, many animal rights activists have called into question the ethics of implanting the device into the brains of innocent animals, many of whom have petitioned to the US government to see an end to Nuerolinks animal testing. The essential question here perhaps is – Is it ever okay to experiment on animals to advance the human condition?

Despite the backlash received from animal rights activists, the video marked an important milestone in neurotechnology, in just one small device capable of receiving and sending brain signals like never before.

The next step for Neurolink is to be able to start clinical trials whereby humans will become the experimental subjects. The N1 is currently awaiting FDA approval before it can be tested on humans. If Neurolink does get accepted for human trials, the implanting of it into the human brain will involve major, invasive neurosurgery that doesn’t come without risk. This type of surgery requires a patient to have a hole drilled into their skull and have the device implanted into the surface of their brain. Infection, bleeding and tissue damage are all common risks of this type of surgery. 

If the clinical trials work and the N1 is successful, the potential to improve patients’ lives who suffer from conditions such as Parkinson’s, epilepsy, dementia and even psychiatric diseases, is abundantly clear, though not without risk.

Will Neurolink eventually succeed in creating a nation of essentially cyborg humans? Will these advances improve human life for the better? Who knows. I suppose we’ll just have to wait and see… 

Are you interested in learning more about AI? Check out Immerse Education’s Artificial Intelligence courses for teenagers here. Spend your summer meeting like-minded peers, advance your skills and knowledge in artificial intelligence and explore one of the world’s most prestigious universities.

Related Content

Tackling homework anxiety: your guide to a calmer study life.