两个团队用 GPT-5.6 相隔三小时解决同一量子密码学难题

2026-08-03 18:49· 23分钟前· Matthias Bastian

跳到正文

AI 摘要

两个独立研究团队分别使用 OpenAI 的 GPT-5.6 Sol Ultra，在相隔仅三小时的时间内解决了同一开放量子密码学问题，并先后将论文提交至 arXiv.org。双方采用了不同方法处理“不可克隆加密”，目前正考虑合并论文。该案例引发思考：当所有研究者都能使用相同模型时，独立发现的标准将如何界定。

原文 · 未翻译

AI is changing how science gets done. Two researchers independently solved the same open quantum cryptography problem using OpenAI's GPT-5.6 Sol Ultra, submitting their papers to arXiv.org just three hours apart, Scientific American reports. MIT PhD student Seyoon Ragavan worked on one side, professors Prabhanjan Ananth (UC Santa Barbara) and Amit Sahai (UCLA) on the other.

Both used the same AI model but took different approaches to "unclonable encryption," a method that relies on quantum properties. They're now considering merging their papers. But the case raises a bigger question: when every researcher has access to the same models, what still counts as an independent discovery?

"If someone mentions an open problem, the first thing is to see if GPT solves it," Ananth says. Ragavan puts it bluntly: "The way I do research now has nothing to do with how I did research two months ago." Math in particular is feeling AI's impact, and reactions range from excitement about new possibilities to depression over a lost sense of identity.