NYU数学教授Tristan Buckmaster与Anthropic数学家Levent Alpöge利用AI模型在重大未解数学难题Navier-Stokes存在性与光滑性问题上取得突破性进展1。该问题是Clay数学研究所设立的七大"千禧年难题"之一，解决者可获得100万美元奖金1。
Buckmaster指控OpenAI在获知他们的工作进展后，迅速集结团队并投入大量计算资源抢先声称解决了该问题1。他指出自己和Alpöge采取的解决路线"几乎没有其他人在做"，质疑OpenAI在数天内通过提示词就能到达同一方向不合理1。Buckmaster声称，在遭质疑后，OpenAI数学研究负责人Sebastian Bubeck要求删除Alpöge的署名，并威胁称"为什么你要毁了你的职业生涯"和"如果你不想我好相处，那我就不必好相处"1。
Buckmaster和Alpöge的研究使用了OpenAI的Codex和Claude两款AI模型1。Buckmaster表示担心其Codex交互数据可能被用于训练OpenAI的模型1。对此，Bubeck否认所有指控，称其为"虚假且煽动性的"，并声称"我是按学术规范来参与讨论的"1。
NYU mathematician Tristan Buckmaster and Anthropic researcher Levent Alpöge achieved a significant breakthrough on the Navier-Stokes existence and smoothness problem, one of seven Millennium Prize Problems set by the Clay Mathematics Institute, using AI models including OpenAI's Codex and Anthropic's Claude.1 The problem's solver stands to receive a one million dollar prize.1
Buckmaster alleges that after OpenAI learned of their research progress, the company rapidly mobilized an entire team and deployed substantial computational resources to pursue the same solution.1 He further contends that OpenAI's head of mathematics research, Sebastian Bubeck, subsequently demanded removal of Alpöge's attribution from the work and made threatening remarks, including "why do you want to destroy your career" and "if you don't want me to be reasonable, then I don't have to be reasonable."1 Buckmaster noted that his specific approach to solving the Clay problem was undertaken by almost no one else, making it unlikely that OpenAI could have reached the same direction through simple prompting techniques within days.1 He has expressed concern that his extensive Codex interactions may have been utilized to train OpenAI's models.1
Bubeck denied these allegations, characterizing them as "false and inflammatory" and stating he engaged "according to academic norms," while expressing disappointment with how events have unfolded.1
评论
还没有评论，欢迎留下第一条。