Mathematics: Proofs, Prizes and AI-Assisted Results — 2026-09-23
A week after OpenAI's Navier-Stokes claim, the mathematical community is digesting the fallout: NPR reports mathematicians "learn a little" from the AI solution, while 25 Fields Medalists continue warning of an "destructive effect" on research. New AI milestones include a crowd-assisted resolution of a decades-old inverse Galois problem, and fresh introspections from Gil Kalai and Steven Strogatz on what AI-driven proofs mean for the discipline.
Mathematics: Proofs, Prizes and AI-Assisted Results — 2026-09-23
Top developments
NPR: mathematicians say they learn "little" from the AI Navier-Stokes solution
In a piece published 22 September, NPR reports that although OpenAI beat humans to solving (a version of) the Navier-Stokes problem, mathematicians say the AI's solution "isn't telling them much" — the proof advances the record but offers little new conceptual understanding of fluid equations. This is the emerging scientific verdict on the most-watched AI math claim of the month.
NYT: history shows why maths isn't "just a game to let AI solve"
Also on 22 September, a New York Times op-ed by applied mathematicians argues that the long, human process of trial and error — not just the solutions themselves — is what historically drove progress, pushing back against the rush to hand open problems to AI models.
OpenAI says 100+ open problems solved in 24 days
Tech Insider reported (22 September) OpenAI's claim that an internal model resolved more than 100 open math problems plus the Navier-Stokes Millennium Prize problem in 24 days, citing roughly 10,000 autonomous AI agents. Independent verification of the full proof remains outstanding, and the mathematics community has demanded an external audit.
AI finds mysterious symmetries, closing a decades-old inverse Galois problem
Scientific American (22 September) reports that mathematicians used AI to find "mysterious symmetries," solving a decades-old problem, with the full set of ~25,000 cases of one "inverse Galois problem" completed by a parallel effort of amateurs and mathematicians. It is a concrete example of AI-assisted enumeration delivering a definitive research-level result.
Gil Kalai: "Be a Grothendieck!" — on AI and mathematics
On 22 September, Gil Kalai published a reflection on what mathematics is in the AI era, part of an ongoing series on the epistemology of machine-generated proofs, invoking a Grothendieckian view of deep conceptual understanding over sheer result-production.
Local view
- France — Les Numériques (20 September): covers the joint declaration by 25 Fields Medalists, including Cédric Villani and Terence Tao, warning three days after the OpenAI announcement about a possible "destructive effect" of AI on research, calling it "un cataclysme" for the community.
- France — Science et Vie (23 September): mathematician Sylvia Serfaty says "c'est la fin d'un monde, celui des mathématiques artisanales," stressing the AI resolved a version of Navier-Stokes, not the most-anticipated formulation.
- Germany/Switzerland — NZZ (22–23 September): "Nach dem KI-Tsunami" describes mathematicians' reactions oscillating between total refusal and pragmatic AI adoption, with "Brandbriefe" (alarm letters) circulating after the proof shock.
Context & numbers
- The OpenAI Navier-Stokes claim involved ~10,000 autonomous AI agents working 88 hours on parts of a problem with a $1 million Millennium Prize attached; the formal Clay Millennium Prize criteria (peer-reviewed, widely accepted publication) remain unmet according to ongoing community debate.
- 25 Fields Medalists signed the alarm letter on AI in mathematics, warning that benchmark races may kill understanding — the largest coordinated response of laureates in the prize's history.
- While formal-verification pipelines do the heavy lifting for deeper archival claims, the OpenAI result has so far not been fully formalized in Lean publicly — a contrast with recent fully machine-checked results like Erdős 550's Lean 4 formalisation on DeepMind's formal-conjectures repository.
On the radar
- Independent verification of the OpenAI Navier-Stokes proof: watch for Lean formalization efforts and any Clay Mathematics Institute statement; German press (ad-hoc-news) reports science bodies now demanding an "unabhängige Prüfung."
- DIE ZEIT reported plagiarism allegations against the unpublished OpenAI model's work — a claim flagged as contested and still developing.
- Université de Lorraine's Institut Élie Cartan will host a public afternoon on the four 2026 Fields laureates (Wang, Deng, Pardon, Tsimerman) on 24 September, 14h–16h.
This content was collected, curated, and summarized entirely by AI — including how and what to gather. It may contain inaccuracies. Crew does not guarantee the accuracy of any information presented here. Always verify facts on your own before acting on them. Crew assumes no legal liability for any consequences arising from reliance on this content.
Powered by
