Keynote Panel State of AI in the Enterprise
summary
Discussion centers on practical enterprise hurdles: poor data quality and missing datasets, 16,000 lost records from a UK COVID Excel example, and the fact that 70–75% prototypes never reach production. Concrete tools and approaches include TensorFlow work, federated learning for privacy-preserving personalization, and pushing for MLops to operationalize models. Real deployments surface sampling bias (Abraham Wald lesson), explainability and monitoring needs, and one insurer buying external data to expand from a narrow membership base to a general market.
AI-written summary · about
chapters (16)
- 0:00Introduction
- 1:28Introductions
- 6:05Machining
- 6:21State of AI in Enterprise
- 12:10Barriers to AI
- 14:23Challenges
- 16:06Summary
- 18:08Data Quality
- 18:54Culture
- 20:41Bias
- 22:22Bias Mitigation
- 23:45Buyers Classification
- 31:25Design Research
- 33:11Sample Bias
- 35:02Recap
- 36:57Importance of significance p value
related talks
01h 01mPanel discussion: Practical and responsible applications of AI in businesses
10mKeynote: The Last Mile Problem: Why AI W...
38mKeynote: Thriving in the Age of AI
58mAI in Reality Fireside Chat: Enterprise AI & Open‑Source Innovation
36mKeynote: AI, ML, Cloud for ML
51mKeynote Panel: Open Innovation in AI - Present an...
50mRealities & Misconceptions about Enterprise Applications of AI/ML and Data
49mKeynote: Operations and AI with Jen Krieger