OpenAI本周发布了722篇数学论文，声称涉及372个未解决问题的重大突破1。部分论文宣称解决了黎曼假设和Birch-Swinnerton-Dyer猜想等千禧年大奖问题1。然而这一大规模发布随即引发数学界广泛争议，已有3篇论文因初等错误被撤回，多篇因错误被修改1。
OpenAI方面声称已对300篇主要论文进行形式化验证1，但专家普遍对论文质量、可信度和AI验证机制表示关注。值得注意的是，ChatGPT将至少一项高调成果评价为"严重幻觉"，认为不应作为证明引用1。此外，OpenAI曾在9月份发表争议性Navier-Stokes问题解决方案，遭指控使用他人数据1。
OpenAI released 722 mathematics papers this week, claiming major breakthroughs on 372 unsolved problems, with some papers purporting to solve millennium prize problems including the Riemann Hypothesis and the Birch-Swinnerton-Dyer Conjecture, each worth $1 million 1. The ambitious release has triggered significant concern within the mathematical community, as multiple papers have already been retracted or revised due to errors 1.
The company stated it has conducted formal verification on 300 of the principal papers 1. However, at least three papers have been withdrawn entirely for elementary errors, while numerous others required corrections 1. Further undermining confidence in the results, ChatGPT itself assessed at least one high-profile achievement as a "serious hallucination" that should not be used as proof 1. The broad scope of claims and rapid revision cycle has left experts questioning the validation processes and reliability of the mathematical results.
The controversy extends beyond the current release. In September, OpenAI published a contentious proposed solution to the Navier-Stokes problem, facing accusations of using others' data without proper attribution 1. As OpenAI pursues a planned stock market listing with a reported valuation target of $1.4 trillion, the credibility issues surrounding this latest mathematics paper release raise broader questions about the company's research quality standards 1.
评论
还没有评论，欢迎留下第一条。