据Business Insider报道，OpenAI CEO山姆·奥特曼（Sam Altman）在斯坦福大学的一次演讲中表示，许多数学家对OpenAI进军该领域的态度，已从最初“我们完了”的悲观，转向接受AI如今在数字与证明研究中无法剥离的角色。他面对的，是数百名对职业前景感到不安的学生。
奥特曼将这一过程比作哀悼：第一天是不敢相信，“我们完了，这个领域结束了，我们让研究生转专业吧”；第三天，可能是愤怒，“OpenAI做这件事太糟糕了”；到了第五天，许多人会说：“我会有一条不同，但希望更好的数学家职业道路。”
奥特曼的乐观判断，与数学界一些人描述的“生存级转变”形成鲜明对比。今年9月，OpenAI突然闯入数学领域，宣布其模型解决了纳维-斯托克斯方程（Navier-Stokes equations）——一个自19世纪以来难倒数学界顶尖人物的著名难题。两名学者一名来自纽约大学，一名来自Anthropic指责OpenAI使用他们聊天记录中的数据帮助破解该问题。还有数学家抱怨，AI模型即便可能确实解决了问题，也没有展示推导过程，留给数学家的是不那么令人高兴的验证工作。
风波并未就此平息。加州理工学院一批担心AI正快速颠覆数学领域、又希望不被边缘化的本科生，发起了一场鼓励参与者用AI解决长期未解问题的“数学马拉松”。超过700名学者签署公开信，谴责该竞赛是在推广AI“slop”（指低质量AI内容）。OpenAI随后退出赞助。主办方此后调整了规则，竞赛定于11月举行。
陶哲轩（Terence Tao）也表达了类似的忧虑。这位加州大学洛杉矶分校的数学天才20岁获得博士学位。他说，数学的“浪漫时代”已经结束，顶尖学者正像他一样适应一种新流程：验证AI的工作成果，而非独自解题。本月OpenAI宣布又解决700多个问题后，陶哲轩说：“有些东西正在失去。”
与此同时，奥特曼站在辩论的另一面，保持着颠覆者的乐观。谈到AI促使企业重组劳动力时，他说：“我不认为会出现就业末日”。但在数学领域，奥特曼认为冲击比其他学科和行业来得更快。“这是我认为我们所有人见过某个领域经历这一周期最快的一次。”
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
According to Business Insider, Open AI CEO Sam Altman, in a speech at Stanford University, stated that the attitude of many mathematicians to Open AI's march in this field had shifted from the pessimism of the initial “we are finished” to accepting the role of AI, which can now be detached from the study of numbers and proof. He is faced with hundreds of students who are worried about their career prospects.
Ottoman compared the process to mourning: on the first day, I couldn't believe, “We're finished, this field is over, let's let the graduate students switch to a profession”; on the third day, perhaps, anger, “Openai did it too bad”; and on the fifth day, many people would say, “I will have a different but hope for a better career for mathematicians.”
Otmann's optimistic judgment contrasts sharply with the “survival transformation” described by some in mathematics. In September of this year, Open AI suddenly broke into the mathematics field and announced that its model solved the problem of Navier-Stokes equations — a well-known figure who has been at the top of the mathematics world since the nineteenth century. Two scholars, one from New York University and one from Anthropic, accused Openai of using data from their chat records to help solve the problem. Mathematicians also complained that the AI model, even if it might have solved the problem, did not show the extrapolation process, leaving the mathematicians with less pleasant validation work.
The wind has not subsided. A group of Caltech undergraduates who fear that AI is rapidly subverting the mathematics field and hope not to be marginalized have launched a “mathematic marathon” to encourage participants to use AI to solve long-standing problems. Over 700 scholars signed an open letter condemning the competition as promoting AI “slop” (meaning low-quality AI content). Open AI then withdrew sponsorship. The sponsors have since adjusted the rules and the competition is scheduled for November.
Terence Tao expressed similar concerns. The Mathematics Genius at UCLA won a Ph.D. at the age of 20. He said that the “romantic age” of mathematics was over and that leading scholars were adapting to a new process like him: validation of AI's work, rather than solving it alone. Following the announcement by Open AI this month that more than 700 more problems had been resolved, Tao Chul-hyun said, “Some things are being lost.”
At the same time, Otmann stands on the other side of the debate, maintaining the optimism of subversives. Referring to AI's motivation for enterprises to restructure their labour force, he said: “I do not think there will be an end to employment”. But in mathematics, Ottman thinks that the impact is faster than in other disciplines and industries. “This is the fastest time that I think we all have ever seen an area experience this cycle.”
Auto-fetched and translated by AI Atlas. All rights belong to the original author.