GoKawiilFormer DeepMind scientist Thore Graepel says LLMs cannot truly reason
Thore Graepel, who helped build AlphaGo, has left Google DeepMind and published an opinion piece arguing that large language models lack genuine reasoning ability. He contrasts this with AlphaGo's 2016 victory over Lee Sedol, which he credits to the system's capacity for creative, reasoned decision-making. Separately, a MIT Technology Review writer reports joining a six-month competition involving roughly 500 participants aiming to reduce their biological age, tracked via a leaderboard.
GoKawiil's interpretation of the reporting above, not reported fact.
Graepel's departure and critique suggest growing skepticism among AI researchers about whether scaling language models alone can produce true reasoning, which could influence where future AI research investment goes. His call for a new architecture drawing on reinforcement-learning systems like AlphaGo implies a potential shift in AI design priorities away from pure LLM scaling. The biological age contest highlights ongoing scientific debate about whether such measurements are reliable enough to meaningfully track anti-aging interventions.
- Thore Graepel, an architect of AlphaGo, has left Google DeepMind
- He argues current LLMs lack real reasoning, unlike AlphaGo's game-playing system
- A separate contest will have ~500 people try to lower their measured biological age over six months
Source: technologyreview.com — Thomas Macaulay, 2026-10-02
Published there as: “The Download: a biological de-aging contest and why LLMs don’t reason”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.