OpenAI Releases 722 Mathematics Papers Without Peer Review
Two reports citing a Phoenix Weekly article say OpenAI released 722 mathematics papers covering hundreds of results in areas including number theory, geometry and theoretical computer science. The reports say 16 problems were directly relevant to Fields Medalist Hugo Duminil-Copin's research and quote him comparing AI to a bulldozer that cleared problems he was working on. They also state that the results had not undergone academic peer review, leaving their reliability unconfirmed.
The reported scale and broad disciplinary coverage make this an unusually large mathematics research batch, but the lack of peer review means its claimed results cannot yet be treated as independently verified.
一位拿过数学界顶级大奖的教授,被AI气哭了。
“周二晚上,我哭了,把女儿吵醒了。”
10月8日,法国数学家雨果·杜米尼尔-科潘在一场线上会议上,讲起了自己的经历。他是2022年菲尔兹奖得主,早已经站在数学研究的顶端。让他落泪的,是OpenAI刚刚公布的一大批数学研究成果。
其中,有16个问题直接关系到他的研究。
在随后的电台采访中,他说AI像推土机一样,把自己正在研究的问题一路推平了。仅用0.1秒,经常打工的人就共情了这位顶级数学家的职场困境。它相当于——