Interview
summary
Defines intelligence = ability to achieve goals efficiently across varied circumstances and calls artificial simply nonbiological, claims supervised learning is effectively solved when labeled data scales (example: driving solved with ~100,000 hours of human data or 100x typical training), identifies transfer of knowledge to new domains as the major unsolved problem, predicts human‑level AI within our lifetimes, and describes current work as building tools to unite a global research community and enable collaboration and hands‑on coding.
AI-written summary · about
chapters (4)
related talks
20mDecision Intelligence: The Next Phase in AI
54mAI: Actual Intelligence
13mArtificial General Intelligence: Why aren't we there yet?
04mFrom big data to human-level artificial intelligence
29mWhat on earth does it mean to build AI software?
39mThe Pocket Guide to Artificial Intelligence
47mThere is no such thing as Artificial Intelligence
41mArtificial Intelligence: Let's Think About Learning