Anthropic interpretability researcher tells Tim O'Reilly: an LLM's world model is readable
mlpowered · x · 2026-09-17
Tim O'Reilly interviewed Emmanuel Ameisen, a researcher on Anthropic's AI interpretability team, going deep on what happens inside an LLM while it processes text. Core claims: prediction demands a world model; that world model is readable; and it is at work in every token. Researchers study activation patterns between layers to see which patterns correspond to particular ideas, and can even intervene by replacing activations with different values to observe how behavior changes. O'Reilly notes the most provocative angle is what studying LLMs might teach us about being human.
More from Research
- Radial gate objective cuts diffusion transformer FID from 16.24 to 11.83 — Junha Park · 2026-11-02
- REA: async brain-cerebellum runtime for robots where policies never wait on the LLM — Pengfei Zhou · 2026-11-02
- Microsoft's MemSteer turns unlabeled user histories into agent memory supervision — MicrosoftResearch · 2026-11-02
- SciGame benchmark uses counterfactual worlds to expose LLM agents' priors vs evidence — Yusheng Zhao · 2026-11-02
- On Terence Tao's blog, Frank Calegari calls on mathematicians to adapt after OpenAI's 700-paper release — Hesamation · 2026-11-02
- Tsinghua & Microsoft's HarnessMM cuts agent token use by up to 87.93% while raising success rate — rohanpaul_ai · 2026-11-02