Matt Turck interviews Richard Socher on RSI, world models, and AI-accelerated science
mattturck · x · 2026-09-11
Matt Turck released his conversation with Richard Socher (You.com) on "When AI Builds AI": recursive self-improvement and scientific progress. Topics include why scientific progress is slowing, whether AI can reintegrate the labyrinth of human knowledge, how LLMs learn biology and proteins, next-token prediction as a world model, whether AI can generate truly original ideas, and simulations for discovery. Available on Spotify, Apple Podcasts, and YouTube.
Related event: Matt Turck Talks Richard Socher on AI Building AI(2 posts)→
More from AGI Musings
- Ex-Googler's dark take: being killed by a Chinese model feels more dignified — moultano · 2026-09-11
- Eric Zelikman Teases New Work: To Understand Us, Models Must Model Us — niloofar_mire · 2026-09-11
- Backyard beer-brewing test shows how little normal people care about AI — LoganGrasby · 2026-09-11
- AI discourse is dominated by people who think math talent plus TC equals the whole human experience — wordgrammer · 2026-09-11
- Anthropic Nears IPO as Its Own Researcher Forecasts >10% AI Extinction Risk — rohanpaul_ai · 2026-09-11
- Bayesian ideology debate: is quantifying uncertainty a virtue when any number is arbitrary? — RexDouglass · 2026-09-11