By the end of this lesson, learners will be able to:
Examine real-world case studies (Netflix, IBM Watson) to see AI’s impact and hurdles.
Recognize common AI limitations like data dependency, interpretability challenges, and scaling costs.
Understand the importance of domain expertise, iterative improvement, and curated data for AI success.
Explore both paid and free resources to deepen their grasp of AI’s achievements and pitfalls.
Key Topics to Cover:
Case Study 1: Netflix Recommendations
How Netflix uses ML models to suggest content (collaborative filtering, user behavior analysis).
Impact: user retention, reduced churn, personalized experience.
Case Study 2: IBM Watson in Healthcare
Overview of Watson’s promise (cancer diagnostics, medical literature analysis).
Hurdles faced (data quality, regulatory issues, domain complexity).
Lessons learned: importance of domain expertise, curated data, and iterative improvement.
Common AI Limitations
Data dependency (quality, bias, quantity).
Overfitting vs. underfitting.
Interpretability challenges (especially in deep learning).
Scaling and maintenance costs (compute power, engineering talent).
Artificial Intelligence (AI) has the power to transform industries, revolutionize business strategies, and improve lives. However, not every AI implementation is a success. While some companies leverage AI to enhance user experiences and streamline operations, others face significant challenges, financial losses, or ethical dilemmas.
🟢 Case study: How Netflix uses data to personalize user experiences
Netflix leverages advanced data analytics to offer a hyper-personalized user experience. By collecting extensive data on user interactions—such as viewing history, search queries, and browsing behavior—Netflix's recommendation algorithm suggests content tailored to individual preferences, enhancing user engagement and satisfaction. This personalized approach has been instrumental in Netflix's success, with over 80% of watched content discovered through recommendations.
How Netflix use AI
Highlight: Understand the different form that AI is used in real cases
Why it’s useful: See all the potential ans application of real word just right now
IBM's Watson for Oncology, initially hailed as a groundbreaking AI system to revolutionize cancer treatment, faced significant challenges leading to its eventual failure. Despite partnerships with leading institutions like Memorial Sloan Kettering Cancer Center, Watson often provided recommendations that were inconsistent with local clinical practices and, in some cases, unsafe. These issues, coupled with declining revenues for IBM Watson Health, culminated in the program's discontinuation in 2023.
AI Case Study Saturday: AI in Healthcare Diagnostics - IBM Watson
Highlight: Understand the challenges and solutions
Why it’s useful: Discover how AI can be useful in healthcare settings
Abtran Case Study: IBM Watson for Cybersecurity
Highlight: A case study of collaboration and innovation between a leading BPO, Abtran and Smarttech247 .
Why it’s useful: Discussing the implementation of cognitive technology to drive operational efficiency in cybersecurity using IBM Watson.
🟢 Two-Minute Case Study - Artificial Intelligence: Sephora's Newest Beauty Advisor
Would you accept skin care recommendations from Siri or Skynet? Maybe not, but you might have been served A.I.-picked recommendations, thanks to Sephora.
As retailers ramp up artificial intelligence adoption to aid everything from customer service to product development, Katie Groody—senior research specialist for Beauty US at Gartner For Marketers—talks about Sephora’s use of A.I. to serve you the personalized site content you want.
The Two-Minute Case Study series presents bite-sized best practices across digital efforts. To learn more, visit www.gartner.com/marketing.
🟢 3 Case Studies of Successful Implementations of AI in Healthcare
Offered by Great Learning, this resource provides free courses on APIs, including RESTful API design, web services, and API integrations, essential for understanding how to interact with various data sources.