DeepMind's Neel Nanda Pushes Back on CoT Critics: Imperfect but Still Our Best Safety Tool
NeelNanda5 · x · 2026-09-04
Responding to @bendee983's criticism of chain-of-thought reasoning, DeepMind safety researcher Neel Nanda argues that "CoT being imperfect" does not mean it isn't the best tool we have or that it isn't adding substantial safety value. Losing CoT as a monitorable reasoning path, he says, would be a tremendous loss — a notable voice in the ongoing debate over whether CoT monitoring is a real safety mechanism.
More from AGI Musings
- Ben Goertzel outlines a general recipe for superintelligence 'basic drives': what it improves vs. protects — bengoertzel · 2026-10-21
- Torch vs Flame: researchers debate whether AI's endgame serves humanity or life itself — danfaggella · 2026-10-21
- 2022 interview resurfaced: AI safety researcher says competence, not consciousness, is the risk — davidmanheim · 2026-10-21
- Silicon Valley is a stock market for people: narrative eventually catches up with substance — karinanguyen · 2026-10-21
- Half of humanity is fed by one century-old reaction — time for AI machines to hunt for more — shyamalanadkat · 2026-10-21
- Agent identity verification will be a mess: restaurants ignore AI emails — alliekmiller · 2026-10-21