Anthropic公司的数学家Levent Alpöge声称在AI模型Claude的帮助下解决了困扰数学界近170年的黎曼猜想 14。该证明目前尚未得到独立验证,且Anthropic官方未对此发表评论或参与宣布过程 1。Alpöge表示,他使用公司内部未公开版本的Claude历时约六周生成了这一证明,完整版本采用Lean形式化验证语言编写,长度超过25万行 4。
针对此次事件引发的争议,斯坦福大学博士后Alessandro Masullo发布了一份完全由人类历时12年完成的广义黎曼猜想证明草稿,以回应AI对数学界的冲击 2。Masullo强调其工作“全是人类的,没有AI”,并写道:“如果这项工作要向那些因当前氛围而感到沮丧的数学家传达任何信息,那就是:我们仍然在这里。我们非常活跃。我们不会离开。” 2。与此同时,包括陶哲轩在内的多位数学家指出,对该AI生成的证明进行验证仍有很长的路要走 4。尽管Alpöge描述与AI协作的过程为“意识流的漫步”,但他承认目前的证明缺乏人类易读的解释 2。此外,一群数学家发表联名信批评AI公司通过宣称重大突破来展示产品能力,认为这种做法损害了数学科学 4。
Levent Alpöge, a mathematician at Anthropic, claims to have solved the Riemann Hypothesis using an unreleased version of the company's AI model, Claude. 14 The proof was generated over approximately six weeks and is written in the Lean formal verification language, spanning more than 250,000 lines of code. 4 As of now, the solution has not undergone independent verification. 1 Alpöge described the process as a "stream-of-consciousness walk," noting that the model appeared to "gallop, dance, and sparkle" near the finish line. 2 Anthropic has not commented on the announcement nor participated in its release. 1 Alpöge acknowledged that the timing of the publication coincides with the company's stock listing, a factor he discussed with CEO Dario Amodei. 4
The mathematical community has responded with caution. Terence Tao stated that while the presence of a human author is a positive aspect, significant work remains for validation. 4 Experts such as Chris Birbeck are currently analyzing the results on Zulip using complex terminology from algebraic geometry and number theory, including Galois representations and automorphic lifts. 2 In response to the impact of AI on mathematics, Stanford professor Ravi Vakil released a draft proof of the Generalized Riemann Hypothesis by postdoctoral researcher Alessandro Masullo. 2 Vakil emphasized that this work is entirely human-produced without AI involvement, representing 12 years of labor. 2 Masullo wrote that if his work conveys any message to mathematicians feeling discouraged by the current atmosphere, it is: "We are still here. We are very active. We are not going away." 2 Additionally, a group of mathematicians issued a joint statement criticizing AI companies for leveraging such claims to showcase product capabilities, arguing that this practice damages the integrity of the mathematical sciences. 4
Alpöge expressed guilt over the lack of a human-readable explanation but stated that delaying publication would have been unbearable due to his lifelong obsession with the problem. 2 This event follows previous unverified claims by OpenAI regarding solutions to parts of the Navier-Stokes equations and other high-level mathematical problems. 4 Separately, Alpöge and NYU mathematician Tristan Buckmaster have released another proof related to the Navier-Stokes equations. 2 Public discussion has extended to platforms like Hacker News, where posts referencing Scott Aaronson’s blog about "stochastic parrots" proving the hypothesis gained limited traction, receiving 15 points and no comments. 3 Media outlets such as The Economist have also reported on the possibility that AI may have solved one of mathematics' most important problems. 5
评论
还没有评论,欢迎留下第一条。