We actually covered this concern, mathematicians are grappling with AI solving problems faster than PhD candidates can build careers on them.
Mathematician Francesco Maggi just questioned OpenAI's 400-proof release. Here's what you need to know.
Maggi posted that OpenAI is about to put roughly 400 AI-generated proofs on a public server. He says experts are still working out how OpenAI's Navier, Stokes example works, and it is not an easy task.
That example goes back to September 8, 2026. OpenAI shared an AI-generated solution to the Navier, Stokes Millennium Prize Problem, with a writeup and a formal proof in Lean. Maggi's point is that until mathematicians digest it, its impact on mathematics is limited. He argues that results have to be understood, connected, challenged and reused. His question is what happens if mathematics is generated faster than mathematicians can absorb it, and what value 400 proofs at once actually have.
The pushback came fast. Some say this is normal, since historians have unread documents and astronomers have unstudied galaxies. Others compare it to GitHub, where the world sorted out how to use code of mixed quality. One reply says AI systems could use the results faster than humans can absorb them. A mathematician estimated OpenAI understands no more than 5% of the solutions. Another noted that about 3,000 subfields means 400 proofs is only 1 or 2 per 1,000 specialists.
The fight sits inside a wider split. On September 11, Terence Tao and 24 other Fields Medalists signed an open letter, "A Severe Misalignment of AI in Mathematics". It says AI company goals and the mathematical community's goals are severely misaligned. The Conjecture Institute replied that AI is an accelerant, not a bane, to mathematical progress.
Key numbers:
- About 400 AI-generated proofs reportedly headed to a public server
- September 8, 2026: OpenAI's Navier, Stokes claim, with a Lean proof
- September 11: open letter signed by 25 Fields Medalists
- Maggi's post drew about 3,200 views in its first hours, per the trending page
Mathematicians now disagree over whether releasing proofs in bulk helps the field or just outpaces human understanding.