OpenAI chief scientist: we lack a theory of what models actually know
TheTuringPost · x · 2026-09-09
An article, "We Don't Know What They Know," explores what models have learned that we don't yet understand. OpenAI chief scientist Jakub Pachocki: "We do not have a satisfactory theory of generalization" — and that gap should affect how fast we develop models.
More from AGI Musings
- The Education of a Doomer: an engineer explains why he turned pessimistic about AI — zetalyrae · 2026-09-09
- Economist analysis: AI has created over 1 million jobs, for now a net job creator — FinanceYF5 · 2026-09-09
- "AI Techbros Only Produce Low-Variance Copies of Researchers," Sparks Debate — MoonL88537 · 2026-09-09
- Claude formalizes Fermat's last theorem: 13M-line verified proof in 11 days — burny_tech · 2026-09-09
- GPT-6 Astra rumored to hit ~11.6-hour p80 task horizon, tracking AI-2027 curve — haider1 · 2026-09-09
- NUS argues universities should pursue 'X + AI', not 'AI first' — ProfChesterman · 2026-09-09