The pace at which artificial intelligence is advancing in the field of mathematics is reshaping external perceptions. Recently, OpenAI publicly released as many as 722 mathematical problem proof manuscripts, covering 17 different mathematical fields, and the overall cost of solving these problems is far lower than in the past, causing a significant stir in the mathematical community.
From a technical and application perspective, although these proofs generated by AI can pass strict machine verification, they expose obvious pain points in practical transformation. Most proofs are difficult to convert into knowledge that humans can easily understand, and in practice, there have been cases where a single symbol error caused multiple related conclusions to fail collectively.
This development has triggered deep concerns within the academic community about the research ecosystem. Mathematicians point out that when AI efficiently produces a massive number of proofs, all the subsequent heavy work of sorting, verifying, and interpreting these results falls on human researchers. A more fundamental issue is that current AI mainly completes problem-solving by logically combining existing knowledge, making it difficult to truly generate new ideas and tools that are at the core of mathematical research.
