Mathematicians Keep Using AI Despite $1 Million Proof Dispute and 4,000-Signature Backlash
Updated
Updated · WIRED · Sep 19
Mathematicians Keep Using AI Despite $1 Million Proof Dispute and 4,000-Signature Backlash
3 articles · Updated · WIRED · Sep 19
Summary
Tristan Buckmaster and other mathematicians say they still rely on OpenAI and ChatGPT even after accusing AI firms of using their work without attribution in high-profile math breakthroughs.
OpenAI amended its Navier-Stokes announcement after investigating Buckmaster’s claim, saying his Codex prompts in the two months before its Sept. 8 paper could not have influenced the system through training.
Andreas Thom said OpenAI also revised an August press release after he flagged overlooked prior work, but he still doubts researchers will ever know whether their ideas fed model outputs.
The dispute has widened into a field-level backlash: 25 Fields medalists signed an open letter, more than 4,000 people backed the Leiden Declaration, and 2,000-plus Caltech affiliates urged an AI math hackathon be suspended.
Even critics say AI’s efficiency gains make retreat unlikely, leaving mathematicians to push for ground rules on attribution, data use and how younger researchers can survive in an AI-shaped discipline.
Why are top mathematicians continuing to rely on advanced AI tools while simultaneously protesting that these systems are swallowing their life's work?
If an AI solves a historic math problem using a researcher's uncredited ideas, who truly deserves the ultimate credit and the prize?
Could the integration of AI in formal proofs eventually erode human mathematical culture, or will it merely redefine academic authorship?