Twenty-five Fields Medalists have signed an open letter criticizing AI companies for threatening intellectual labor. The mathematicians argue that the race to solve complex mathematical problems using AI risks undermining academic integrity and the established chain of knowledge transfer.
Tristan Buckmaster, a professor at New York University, accused OpenAI of pressuring him to exclude a colleague from Anthropic as a co-author on a joint research paper. He also raised concerns that OpenAI might have utilized materials uploaded to its Codex platform to derive its own mathematical proofs.
OpenAI withdrew its sponsorship of a mathematics event at the California Institute of Technology following criticism from researchers at the institution. Meanwhile, the company's recent mathematical proof has yet to undergo independent verification by the scientific community.
The letter highlights fears that AI labs may invest tens of millions of dollars into large language models to preempt original researchers. This competitive dynamic threatens the culture of open science, potentially forcing mathematicians to adopt greater secrecy to protect their findings.
The signatories emphasize that AI-generated solutions must be understandable and explainable to the broader community. The document follows the Leiden Declaration published in June, which addressed the impact of large language models on mathematical research and provided recommendations for institutions.
