AI’s advances in mathematics prompt a debate over the future of mathematicians

Summary

At the 2026 International Congress of Mathematicians in Philadelphia, researchers confronted AI systems’ rapid progress on mathematical problems. The article reports that Anthropic’s Claude Fable produced a counterexample to the Jacobian conjecture, while ChatGPT had recently solved two other open problems; it also says mathematicians used AI to solve a Millennium Prize Problem. These developments have intensified concern about whether human researchers will retain a role and whether early-career mathematicians will face fewer academic jobs. Terence Tao called the moment a crisis in mathematics’ values and practices, but argued that mathematicians should not simply reject AI. He urged them to establish and enforce their own standards for acceptable use, rather than let companies set the rules, and praised the Leiden Declaration’s effort to develop shared norms. One disputed proposal is that AI models should not be credited as authors of proofs. Views across the field remain divided. Some researchers expect AI to make mathematics more productive and see human work in interpreting results; others question whether AI companies’ incentives align with mathematical values and advocate refusing or limiting the technology. Mathematicians also worry that AI could reduce jobs, widen disparities between well-connected researchers and those with less security, and weaken students’ experience of struggling with difficult problems. The article emphasizes that even if AI can produce results, understanding and finding meaning in them may remain valuable human work. It leaves unresolved whether people will continue to understand major AI-generated proofs, and how the next generation should be trained for a field increasingly shaped by these systems.