Machine Learning Capabilities for Applications
A talk about AI capabilities and pitfalls for non-specialists
If you are an AI non-specialist or you want something to help get a non-specialist up to speed on the big picture, here is a half-hour talk (plus Q&A) about different flavors of machine learning-based AI, including key application capabilities, what it can do well, and what it can’t.
YouTube video Presented at Business of Semiconductor Summit 2024, Sept. 11, 2024. (Includes both the talk and a Q&A session.)
Abstract:
While most discussions of machine learning concentrate on the underlying mechanics, this talk discusses their capabilities, strengths, and weaknesses at the level of how they can be applied to a wide variety of real-world applications. Capabilities discussed are classification, end-to-end behavior, generative outputs, and foundation model applications. Challenges discussed include bias, validation, edge cases, hallucinations, autonowashing, AI safety, and accountability. The evergreen concept of the 90/10 principle cuts both ways for AI, with solving the last 10% of building dependable systems likely to make the difference between winning and losing technology bets.
The pesentation is clear and to the point. But I must say I really enjoyed the Q&A session with Phil Koopman. It's worth watching.