Ex-DeepMind researcher argues LLMs still lack auditable reasoning, calls for AlphaGo-style rethink
ThoreG · x · 2026-10-02
Former Google DeepMind researcher Thore Graepel marks the 10th anniversary of AlphaGo's Move 37 with an MIT Technology Review essay arguing that today's AI systems are missing something fundamental.
Key points:
- Move 37 wasn't intuition alone: AlphaGo could search possible futures, test its instincts, and reason about outcomes
- LLMs are remarkably capable, but longer chains of thought are not genuine reasoning — they typically lack an explicit, inspectable record of what they know, what remains uncertain, what evidence supports a conclusion, or whether real progress has been made
- This is why he recently left DeepMind, advocating a fresh approach to machine reasoning drawing on AlphaGo's architectural lessons
- For AI to produce trustworthy, genuinely novel insights in science and medicine, conclusions must arise from an auditable process of evidence, inference, and belief revision
Related event: AlphaGo Veteran Questions LLM Reasoning, Sparking Researcher Debate(6 posts)→
More from AGI Musings
- Dev challenges "AI will solve humanity's problems": where exactly in the next 5 years? — aronchick · 2026-11-17
- Top researchers are quietly stopping paper submissions, observer notes — avt_im · 2026-11-17
- "Abundant intelligence should terrify businesses built on keeping it scarce" — _AustinCalvert_ · 2026-11-17
- 130-page paper introduces "bottleneck pressure": work AI can't do gains value as the rest speeds up — soumitrashukla9 · 2026-11-17
- AI conference authors and reviewers complain unprecedentedly about AI slop — khademinori · 2026-11-17
- Sam Altman: AI already speeds up OpenAI's research, but a self-improving button? 'Definitely not' — ChrisGPT · 2026-11-17