AI’s progress on difficult mathematical problems is prompting researchers to debate not only what the technology can solve, but what it could change about mathematical work itself. A Times Higher Education feature describes concerns over research, teaching and attribution, alongside mathematicians who see AI as a useful tool.
Watch Desk analysis
What happened
About 300 scholars attended the Royal Society’s “Mathematics in the age of AI” event on 29 October, with another 700 joining online, the article says. It follows an open letter calling for an AI “emergency”, signed by 42 Royal Society fellows and foreign members and later endorsed by another 500 mathematicians.
The letter points to recent advances by leading AI models and an OpenAI model’s claimed solution to the Navier-Stokes problem. The feature notes controversy over how that problem was solved. It also says OpenAI released 722 manuscripts claiming solutions to other open mathematical problems at the start of October. OpenAI denies that prompts by mathematician Tristan Buckmaster influenced its system; Buckmaster has claimed his work may have been leaked and used to inform prompts, according to the article.
Why it matters
The questions go beyond whether a model can produce a proof. Mathematicians quoted in the feature worry that fast, poorly explained results could make it harder to check credit and prior work, while changing how students learn to do research. One concern is that using AI too heavily could leave junior researchers without skills they would otherwise develop.
There is also a clash over incentives: companies may present famous problems as demonstrations of model capability, while mathematicians value careful explanation, shared ideas and training. The feature reports that some researchers are optimistic about AI as a research tool, so this is not a simple vote for or against machines in mathematics.
Our read
The interesting question is not whether AI belongs in mathematics. It already does. The question is what counts as a useful mathematical result: a striking answer, or work that can be checked, understood and built on by other people. If companies want the field’s applause, they will need more than a headline-grabbing solution.
What to watch
- Whether companies publish proofs and methods that mathematicians can independently examine.
- How universities and supervisors advise students on using AI in research.
- Whether mathematicians develop shared expectations for attribution and disclosure.
Discussion spark: Should mathematical AI breakthroughs be judged mainly by whether they solve hard problems, or by whether their methods are transparent and useful to the wider research community?
Sources and evidence
- Source update (19 November 2026, 00:00 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.