This September, the artificial intelligence sector shocked the academic world. OpenAI claimed it resolved the Navier-Stokes existence and smoothness problem.
This notorious fluid dynamics puzzle is one of seven Millennium Prize Problems. Instead of universal applause, the announcement sparked a fierce academic backlash.
Twenty-four Fields Medal winners, including Terence Tao, signed an open letter condemning OpenAI’s tactics. The dispute extends far beyond a single mathematical proof. It exposes a growing rift between human scientific pursuit and brute-force corporate AI.
The 88-Hour Brute Force Controversy
The conflict stems from how OpenAI secured its discovery. NYU mathematics professor Tristan Buckmaster and his collaborator Levent Alpöge, a mathematician currently working at Anthropic, had spent nearly a year developing a solution for the Navier-Stokes problem.
They were on the verge of publishing their proof which demonstrates that fluid dynamics can develop a finite-time singularity, or “blowup” when rumors of their progress circulated within the academic community.
OpenAI caught wind of these rumors on September 1st. In response, the company deployed an unreleased, highly advanced internal model reportedly a successor to GPT-6 Astra launching 10,000 AI agents to attack the Millennium Prize Problems simultaneously.
Working continuously, the swarm generated 2.7 million messages and consumed roughly 130 billion output tokens. In just 88 hours, the agents successfully generated an analytical proof and a formalized verification in Lean, effectively scooping Buckmaster and Alpöge.
The mathematical community was appalled by the subsequent corporate maneuvering. OpenAI reportedly offered Buckmaster a joint announcement but outright refused to include Alpöge due to his employment at rival Anthropic. Allegations soon surfaced that OpenAI’s researchers pressured Buckmaster to sever ties with his colleague.
Beyond the interpersonal drama, the incident proved that tech giants can now weaponize immense computational resources to snipe research from independent academics.
OpenAI did not invent the initial conceptual framework; they simply outcomputed the human mathematicians the moment they sensed a breakthrough was near.
The Threat to Open Science and Conceptual Insight
The open letter signed by the Fields Medalists warns that OpenAI’s approach could permanently damage the foundations of mathematics.
Terence Tao highlighted a chilling new reality: if the mere whisper of a promising research direction can trigger a massive, AI-powered effort to flatten the problem, researchers will stop sharing their ideas. Centuries of open scientific tradition rely on collaboration and early peer discussion.
That collaborative spirit cannot survive if tech giants use their financial and computational supremacy to instantly monopolize any leaked mathematical insight.
Furthermore, the signatories argue that AI companies are treating mathematical proofs as mere benchmarking metrics rather than intellectual pursuits. Solving a Millennium Problem is traditionally valued not just for the final true-or-false answer, but for the new analytical tools, frameworks, and human insights developed along the way.
A machine that grinds out a verified but highly convoluted Lean proof over a weekend skips this vital intellectual journey.
If artificial intelligence merely churns out complex resolutions that humans cannot intuitively grasp, the ancillary benefits of mathematical discovery are lost. The scholars emphasize that optimization and problem-solving are proxies for conceptual understanding, not replacements for it.
If the industry normalizes deploying algorithmic dragnets to hijack human research, humanity might acquire a library of infallible proofs while slowly losing the ability to understand the universe they describe.
Source: The Economist, "Top Mathematicians Are Outraged by OpenAI’s Methods"
