前 DeepMind 研究员创办 Metis Reasoning，想把 AlphaGo 搜索原理用于消除幻觉
Former Deepmind researcher thinks AlphaGo's playbook can cure AI hallucinations
DeepMind 出走的研究员办了家新公司，想法挺有意思：让 AI 像下围棋那样一步步推，结论都有据可查。不过现在还只是纸上谈兵。
前 DeepMind 研究员 Thore Graepel 创办初创公司 Metis Reasoning，希望把 AlphaGo 的搜索原理迁移到通用推理任务上。方案的核心是用一个显式知识账本（knowledge ledger）替代把知识锁在模型权重里的做法，让每个结论都能对照证据逐条检查。目前公司还没有可运行的系统，也没有公布任何基准测试成绩。
Former Deepmind researcher thinks AlphaGo's playbook can cure AI hallucinations
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. The article Former Deepmind researcher thinks AlphaGo's playbook can cure AI hallucinations appeared first on The Decoder .
