Ex-DeepMind researcher proposes AlphaGo-style ledger against hallucinations
Original title: Former Deepmind researcher thinks AlphaGo's playbook can cure AI hallucinations
Metis Reasoning’s white paper is only a research program, with no working system, benchmarks, or model comparisons.
Thore Graepel, a former DeepMind researcher who worked on AlphaGo, AlphaZero, and MuZero, founded Metis Reasoning and outlined the approach at Nvidia’s GTC in Berlin. The company wants an explicit knowledge ledger, independent checks, and search that can abstain, arguing larger language models will not stop fabricated answers. The paper acknowledges partial real-world information and costly evidence, and Bloomberg reports Graepel is raising funding.
Unconfirmed: Fundraising figures are attributed to Bloomberg, and the paper has no demonstrated system.
- Graepel left Google DeepMind’s post-AGI team over the summer.
- The ledger would record known facts, confidence, and open questions.
- Bloomberg says he is raising a low tens-of-millions round.
- He also holds the machine learning chair at University College London.
- AI models
- AlphaGo
From the original article
4 min read at the-decoder.com
Former Deepmind researcher Thore Graepel wants his startup Metis Reasoning to bring AlphaGo's search principle to general reasoning. Instead of locking knowledge inside model weights, an explicit knowledge ledger would make every conclusion checkable against evidence. So far, though, there's no working system and no benchmarks.
By Jonathan Kemper · the-decoder.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.