Mathematicians warn AI problem-solving harms the field
Original: A misalignment of AI in mathematics
Why This Matters
Mathematics is both a bellwether for AI capability claims and a foundation for future technology — if the field's knowledge pipeline degrades, downstream effects could take decades to surface.
A growing group of mathematicians has published a declaration arguing that AI companies' push to solve famous math problems as benchmarks is misaligned with the actual goals of mathematics: conceptual understanding, idea development, and human transmission of knowledge across generations.
The declaration, posted at mathandai.org and signed by an expanding list of endorsers, argues that while LLMs have recently demonstrated the ability to solve major open problems across multiple areas of mathematics, framing this as progress is misleading and potentially damaging. The signatories draw a sharp line between solving a problem and understanding it. In mathematics, a solved problem is typically the beginning of a long communal process: talks, discussions, simplification, and eventually a textbook treatment accessible to students. AI-generated solutions, they argue, short-circuit this entirely. The declaration is particularly pointed about pace and attribution. Solutions are 'announced in a rush, leaving no time for a proper writeup, the isolation of new methods and ideas, and citing relevant previous work.' This raises what the authors call 'severe attribution and plagiarism questions.' They also warn that without mathematicians willing to integrate AI-conceived ideas into the canon, those ideas will never fully live — the human transmission chain will break. The signatories situate the math problem within a broader concern: AI systems trained on accumulated human work are now capable of producing outputs that traditionally required years of skill-building, threatening the very processes that generate new questions and ideas, not just answers. The declaration explicitly notes it sees these issues as part of wider misalignment affecting other scientific and creative professions.