Many mathematicians are grappling with OpenAI's announcement on 6 October, and this community snapshot captures their gut reactions. Posted on 8 October 2026 and updated on 9 October 2026, the collection features reflections from contributors who are signed only by name, without reference to affiliation or career stage. Readers will find both general reflections and problem-specific comments.
Tim Santens called the results of OpenAI almost unbelievable, noting that Quasi-Riemann was a result he had considered complete science fiction. He argued that mathematics as a research discipline will have to undergo a massive transformation given the existence of these models.
Roman Sauer described being stunned by the solutions to problems 207 and 335. He said a mathematical technique he had relied on in a paper on macroscopic scalar curvature with Sabine Braun, later extended by Hannah Alpert, now appeared in OpenAI's creative solutions to two very different problems. He concluded that mathematicians must radically expand their view of what a mathematician is and does.
Xiaolei Wu, writing on the last day of a national holiday in China, recalled receiving the same PDF file on their phone and finding that OpenAI had apparently solved many open problems in group theory. Among the results were that Thompson's group F is non-amenable and that the Eilenberg–Ganea conjecture is false, while the Whitehead Conjecture remained standing. The Borel conjecture failing in dimension 4 was another striking result.
Nicholas Williams observed that the release taught the community many new facts, but said mathematicians will need time to digest the papers. He stressed that the end goal must always be human understanding, and that mathematicians do more than solve open problems from the literature.
Source: Hacker News · Summarized by HeadlinesBriefing