大模型
x.com ·
@GaryMarcus
Gary Marcus 转发 Thore Graepel：LLM 长思维链不等于真正推理，需可审计的推理过程
Gary Marcus 转发 Thore Graepel 的文章观点：AlphaGo 的 Move 37 来自可搜索、可检验的推理过程，而当前 LLM 缺乏对已知、未知、证据与进展的显式可检查记录，生成更长的思维链并不等于真正推理；Graepel 称因此离开 Google DeepMind，主张借鉴 AlphaGo 架构重新设计机器推理。
AI推荐理由 引用前 DeepMind 研究者 Thore Graepel 的观点：LLM 生成更长的思维链不等于真正推理，缺乏可审计的知识、不确定性与证据记录，并称其因此离开 Google DeepMind。涉及 AI 推理架构的核心争议，有具体人物与立场，但为观点性内容，无技术细节或实验证据。
70
浏览 2,421
点赞 32
评论 1
转发 11
正文
💯💯💯: “If AI is going to produce trustworthy and genuinely novel insights in science, medicine and beyond, we need systems whose conclusions arise from an auditable process of evidence, inference and belief revision.”
It is madness to believe otherwise.
Thore Graepel@ThoreG
Ten years ago, AlphaGo’s Move 37 shocked the world.
It wasn’t intuition alone that produced it. AlphaGo could search possible futures, test its instincts and reason about what would happen next.
In a new piece for [@techreview](https://x.com/techreview), I argue that today’s most advanced AI systems are still missing something fundamental.
LLMs are remarkably capable, but generating longer chains of thought is not the same as genuine reasoning. They typically have no explicit, inspectable record of what they know, what remains uncertain, what evidence supports a conclusion or whether genuine progress has been made.
This is why I recently left [@GoogleDeepMind](https://x.com/GoogleDeepMind). I believe we need a fresh approach to machine reasoning, drawing on some of the architectural lessons from AlphaGo.
If AI is going to produce trustworthy and genuinely novel insights in science, medicine and beyond, we need systems whose conclusions arise from an auditable process of evidence, inference and belief revision.
[🔗 View Quoted Tweet](https://x.com/ThoreG/status/2105988456824386034)