The Future of Mathematics: How Mathematicians Should Pursue Mathematical Understanding in the AI Era

Against the backdrop that AI can now easily generate results that were previously publishable, mathematician Jeremy Avigad proposes three paths for mathematics in the AI era: solving harder problems, thinking about bigger problems, and redefining mathematical understanding itself. He points out that although current foundation models have read all mathematical literature and are tireless, they rely on reinforcement learning to piece together existing techniques and still lack creativity and high-level strategy.