ReadingResearchRadarInvestment frameworkAI Company Primary Sources
Sign in / Sign up
← Back to reading
TopicDisconfirmAI EditorialUpdated 2026-10-28 19:03

ETH AI证明争议

4 steps1 pieces4 Sources

2026年10月,OpenAI和Anthropic分别发布由AI模型生成的数学证明,声称推翻了自1999年以来主导理论计算机科学的指数时间假设(ETH)。这一发现若被证实,将动摇数百篇依赖ETH的论文基础,并轻微提升P=NP的可能性,但并未直接解决该问题或威胁现有加密系统。

【背景】ETH认为3-SAT等逻辑难题无法以亚指数时间求解。OpenAI称其模型于10月16日发现反例,并于次日通过Lean形式化验证;Anthropic随后也发布了基于不同方法的类似结果。

然而,科学界反应复杂。MIT教授Ryan Williams与ETH提出者Impagliazzo、Paturi组成工作组进行审查,但他们在公开信中强调:“我们的参与不应被视为对该主张的认证。”Williams指出,虽然部分证明步骤看似常规,但存在“外星般陌生”的关键环节,需社区共同解析。此外,OpenAI此前因“DDOS式”批量发布证明而受批评,此次虽调整了流程(如资助会议、公开原始数据),但关于AI贡献规范及验证透明度的争论仍在持续。目前,尚无外部人类专家确认这两份证明的有效性。

How it got here

  1. 2026-10-16

    OpenAI声称其内部AI模型发现了ETH的反例算法;同日,Anthropic声称其Claude模型也独立发现了相同结果。

    sansser.com ↗
  2. 2026-10-17

    OpenAI将ETH反驳证明在Lean中完成形式化验证,并将相关文件和GitHub仓库公开。

    skimwire.com ↗
  3. 2026-10-26

    OpenAI正式发布公告,宣布AI解决了包括ETH在内的三个重大数学/物理问题;数小时后,Anthropic发布由Claude撰写的论文,声称以不同方法同样推翻了ETH。

    techcrunch.com ↗
  4. 2026-10-28

    TechCrunch报道MIT教授Ryan Williams接受采访,解释社区正在组织的验证工作,并指出这“大概不是最终答案”。

    techcrunch.com ↗
Sign in / Sign up
ReadingResearchRadarInvestment frameworkAI Company Primary Sources

What is still unsettled

  • 外部复杂性理论家能否成功复现并确认OpenAI和Anthropic提供的Lean形式化证明?
  • 针对AI生成科学发现的验证规范和伦理标准,社区将在即将召开的会议上达成何种共识?
  • 如果ETH被确证为假,哪些依赖其下界证明的具体子领域(如细粒度复杂性)需要立即重写?

Sources

  • techcrunch.comTechCrunch: Frontier AI labs say their models refuted the Exponential Time Hypothesis↗
  • sansser.comSansser: Two AIs just broke a 27-year-old rule of computer science↗
  • skimwire.comSkimwire: OpenAI says its AI disproved the Exponential Time Hypothesis↗
  • decrypt.coDecrypt: OpenAI Proves ETH Is False, and Crypto Twitter Has Questions↗

Coverage in this topic

  • MIT Professor of Computational Complexity Ryan Williams: AI Claims to Refute Exponential Time Hypothesis, Credibility of P=NP Rises2026-10-28News

    OpenAI and Anthropic have respectively released AI-generated mathematical proofs claiming to overturn the long-held belief in computer science regarding the Exponential Time Hypothesis (ETH). MIT professor of computational complexity Ryan Williams explains the implications of this discovery for the P versus NP problem, the verification efforts being organized by the community, and why he believes this is 'probably not the final answer.'