AI is breaking our proxies for expertise

Mathematicians worry that AI's success in solving prestigious problems threatens the field's core purpose of generating conceptual insights, not just finding solutions. While problem-solving serves as a proxy for mathematical progress and prestige, the real intellectual work lies in developing new ideas and concepts—and mass-producing AI-generated solutions could undermine this foundational work by divorcing problem-solving from the deeper understanding it should catalyze. Nearly 5,000 mathematicians, including 25 Fields medalists, have signed a declaration expressing concern that prioritizing AI's problem-solving capabilities over idea-generation could damage mathematics' ability to develop meaningful new knowledge.