Mathematicians Fear AI's Dominance Yet Depend on Its Tools
A prominent mathematician claims OpenAI leveraged his unpublished work to solve a long-standing problem, yet he continues using the company's AI tools because they are indispensable. The incident has sparked debate over whether AI will render human mathematicians obsolete, while researchers acknowledge they have little choice but to rely on these powerful models. Despite the existential threat to the field, many mathematicians find the utility of AI too great to abandon.
The Navier-Stokes problem carries a $1 million bounty and has stood as one of mathematics' most celebrated open questions. Buckmaster worked on it alongside an Anthropic researcher, using both OpenAI's Codex and Anthropic's Claude. OpenAI later deployed tens of thousands of agents to reach a solution, prompting Buckmaster to allege the company capitalized on his near-breakthrough without proper acknowledgment.
German mathematician Andreas Thom similarly discovered OpenAI's Astra model had proven a problem he'd spent two decades developing techniques to solve. Both researchers received amended statements from OpenAI, yet neither received definitive answers about whether their private work influenced the systems. The incidents highlight how AI's opacity is dismantling traditional academic credit systems, leaving mathematicians unable to verify contributions or trace intellectual lineage.
This story could reshape how academic research is conducted and credited. If mathematicians cannot trust that their unpublished work remains protected from AI training pipelines, they may become more secretive, slowing collaborative progress. Conversely, the demonstrated utility of these tools may force institutions to develop new frameworks for attribution and consent. Early-career researchers could face particular pressure, caught between career incentives to publish and fears their work will be absorbed without recognition. The outcome may influence how other knowledge-based professions negotiate their own relationships with powerful AI systems.