数学研究社区与OpenAI的合作关系因多项质量和伦理问题而告终。2026年9月，OpenAI宣称利用人工智能解决了纳维-斯托克斯千禧年难题，投入约1500万美元的token费用 1。包括菲尔兹奖得主特伦斯·陶在内的25位国际顶尖数学家随后发表公开信《AI在数学中的严重错位》，指出这一成果存在数据伦理问题、证明难以阅读且属性不清等关键缺陷 1。
OpenAI随后于10月6日发布了700多个文件，声称其中包含多个数学难题的解决方案 1。然而学术界对这些成果的真实性和学术规范提出了严重质疑。数学家Tristan Buckmaster指出，这些证明中关于Boussinesq和Euler方程的写作"更接近模型在人类指导下的产物，而非真人撰写的论文" 1。面对这一困局，人类数学协会在10月7日发表声明，呼吁数学家停止与OpenAI的合作 1。
这场冲突的背景可以追溯到2023年10月，当特伦斯·陶开始使用ChatGPT学习Lean证明助手时，许多人对AI与数学的融合抱有希望 1。然而，OpenAI随后的行动被普遍认为违反了学术规范，损害了学术界对AI工具的信任。
The mathematics research community has formally broken with OpenAI following disputes over the company's claims to have solved the Navier-Stokes Millennium Prize Problem. In September 2026, OpenAI announced it had used artificial intelligence to address the long-standing mathematical challenge, spending approximately $15 million in token costs to do so 1. However, the approach drew sharp criticism from leading mathematicians over data ethics violations, unreadable proofs, and unclear attribution 1.
The conflict escalated when more than 25 Fields Medal recipients, including renowned mathematician Terence Tao, published an open letter in September 2026 titled "Critical Misalignments of AI in Mathematics," accusing OpenAI of disregarding established academic standards 1. Tristan Buckmaster noted that the proofs generated resembled output produced under human guidance rather than original human scholarship, stating that the various Boussinesq and Euler formulations "were closer to models produced under human direction than papers written by actual mathematicians" 1. Following OpenAI's release of over 700 files on October 6, 2026, purportedly containing solutions to multiple mathematical problems, the newly formed Human Mathematics Association issued a statement on October 7, 2026, urging mathematicians to cease collaboration with OpenAI 1.
This rupture marks a significant moment in the relationship between artificial intelligence and academic research 1.
评论
还没有评论，欢迎留下第一条。