AI systems are producing results that mathematicians say would have seemed unimaginable only months ago. The change is startling, but leading mathematicians interviewed by The Telegraph describe more than a profession being bulldozed: they see new ways to check proofs and pursue problems, alongside real questions about what people should do next.
Watch Desk analysis
What happened
Fields Medal winner Hugo Duminil-Copin tells The Telegraph that he first felt AI’s advances as a threat to his research, but now hopes mathematics can adapt and benefit. Oxford professor and Clay Mathematics Institute president Martin Bridson called recent results “absolutely breathtaking”, while stressing that claimed solutions to Millennium Prize Problems still need checking. The institute’s review of the Navier-Stokes claim could take months.
The article describes two striking examples. OpenAI said in September that it had solved the Navier-Stokes existence and smoothness problem, using more than 10,000 AI agents over 88 hours, according to the report. Anthropic said a high-end Claude model completed a formalisation of Fermat’s Last Theorem in 11 days. Imperial College mathematician Kevin Buzzard says his team is now working to interpret and refine the AI-produced Lean proof. These are reported claims and accounts, not a declaration that the mathematical community has settled every question.
Why it matters
The immediate change is not simply that a machine can announce a proof. AI can generate candidate work at speed, while formal verification tools such as Lean can help check whether a proof follows logically. Researchers still have to understand what the system has done, assess the result and decide what it means. The Clay institute’s deliberately unhurried review is a useful reminder that a dramatic result and an accepted solution are not the same thing.
There is also a question of what gets lost or gained when a long mathematical chase becomes faster. Bridson argues that AI could shorten the path from fundamental theory to practical applications. Buzzard, meanwhile, says automation can free mathematicians from laborious formalisation work. Neither outcome is automatic; both depend on whether the new tools produce work people can inspect and build on.
Our read
This is a consequential shift, not a neat tale of machines replacing mathematicians or saving them. The best detail is the human work continuing after the headline result: researchers are checking, interpreting and improving proofs. That is less cinematic than an AI solving a centuries-old problem in a weekend, but rather more useful.
What to watch
- Whether the Clay Mathematics Institute accepts the Navier-Stokes claim after review.
- How mathematicians check and refine AI-generated proofs, including in Lean.
- Whether faster proof work leads to new results or practical applications, not just faster announcements.
Discussion spark: If AI can produce a proof faster than mathematicians can understand it, should the result count as progress before people can explain and verify it?
Sources and evidence
- Can our anguished mathematicians overcome AI dread? – The Telegraph (13 November 2026, 06:00 UTC)
- Malawi hails British Empire as ‘liberators’ who saved Africans from slavery – The Telegraph (13 November 2026, 06:00 UTC)
- Navy buys three mine-hunting ships to protect North Sea from Russia – The Telegraph (13 November 2026, 06:00 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.