Reactions to 100 Solutions

A wide-ranging compilation of over a hundred short reactions from mathematicians responds to OpenAI’s October announcement that its models produced solutions to many long‑standing open problems. The collection documents specific claimed breakthroughs - Quasi‑Riemann, Thompson’s group F shown non‑amenable, Eilenberg-Ganea false, Boone-Higman solved, the Borel conjecture failing in dimension four, failures of the coarse Baum-Connes conjecture, and other high‑profile targets such as the Cannon conjecture and non‑residually‑finite hyperbolic groups - and notes surprising technical overlaps (for example, ergodic‑theory constructions reused across distant problems). Contributors emphasize that the released write‑ups vary wildly in quality, were produced as part of OpenAI’s internal model evaluation using unsolved problems, and were published without named authorship, standard peer review, or transparent documentation of methods and failure rates.

The reactions cluster around two themes: exhilaration at new mathematical content and deep alarm about institutional and ethical consequences. Many call for rapid community action to verify, digest and curate the results on a neutral platform, and to demand transparency and attribution. Others warn that the dump of solutions will erode norms that structure mathematical progress, damage career incentives for trainees, and concentrate control in corporate hands, while some remain intent on mining new techniques from the claims. Overall the compilation portrays a discipline forced to confront a sudden, disruptive shift in how problems get solved, communicated and validated.