Dietterich: We must test if agent behavior matches their language, not just assume CoT implies causality
tdietterich · x · 2026-09-01
Thomas Dietterich commented on the need to verify if agent behavior aligns with their language descriptions. He suggested intervening on the language to prove causality, noting that Chain of Thought often does not predict behavior accurately.
Related event: Researchers Question Whether Agent Behavior Matches Language(2 posts)→
More from AGI Musings
- Sam Altman Predicts Internal AGI by Year-End; Anthropic Targets 2027 — haider1 · 2026-09-01
- Opinion: Coding Harnesses Are Mostly Solved, While Others Have Barely Begun — tobowers · 2026-09-01
- Anthropomorphism vs. Anthropomimesis in AI Design — teortaxesTex · 2026-09-01
- Americans Should Care More About Being Beaten by China, Not Just Europe — NathanpmYoung · 2026-09-01
- AI Consciousness Paradox: We Infer It in Humans but Ask AI Itself — robleclerc · 2026-09-01
- Wilbur Wright Predicted Humans Would Not Fly for 50 Years — PeterDiamandis · 2026-09-01