Top mathematicians fear AI may end their field — and it's already making breakthroughs

In early August, about 40 leading mathematicians gathered at OpenAI's office in San Francisco to discuss the future of their profession amid the rapid advancement of artificial intelligence, as reported by The Washington Post. The central question of the meeting: what will be left for humans to do if AI surpasses them in research-level mathematics in the near future?

University of Toronto professor Daniel Litt even titled his presentation "The End of Mathematics." He outlined a worst-case scenario in which AI generates cutting-edge mathematical research in such volume that the need for human mathematicians sharply declines. However, Litt himself considers this scenario unlikely but stresses that the field must adapt now.

The mathematicians' concerns are fueled by real AI achievements. In May, OpenAI presented a result that many experts called the first major mathematical breakthrough achieved by AI. One of the company's unreleased models found a counterexample to a long-standing conjecture about unit distances. Mathematician Jacob Zimmerman stated he would accept this result for publication in any scientific journal without hesitation.

Since then, OpenAI and Anthropic have published a number of other mathematical results. Some of them bypass the traditional scientific workflow, where discoveries are first discussed at seminars, verified by peers, and only then published in journals. This raises a separate concern: mathematicians fear that a massive influx of AI-generated results will make it difficult to determine what has actually been proven.

In June, over 3,000 mathematicians signed the Leiden Declaration, calling for the responsible use of AI in science. The declaration emphasizes the need for transparency and verification standards when AI contributes to research. The gathering at OpenAI and the subsequent discussions highlight a growing unease: the very tools that could accelerate mathematical discovery may also undermine the profession's traditional structures and the trust that underpins scientific validation.