OpenAI’s Speedy Math Victory Raises Ethics Questions in Academia

OpenAI’s announcement Tuesday that it solved a Millennium Prize problem, specifically the Navier-Stokes equation, represents both a significant achievement and a stark reminder of how quickly AI is transforming mathematics. However, the breakthrough is complicated by allegations of speed-running to beat other researchers, sparking fears that the company’s actions could damage the field's long-standing norms.

According to Abhishek Saha, a mathematics professor at Queen Mary University of London, OpenAI engaged in "the kind of things that mathematicians will generally not do."

In a blog post published Tuesday, OpenAI revealed that its unreleased model solved the Navier-Stokes problem in just 88 hours. The problem, which carries a $1 million bounty, has stumped researchers for 90 years. The company utilized a swarm of approximately 10,000 AI agents powered by its internal model, calling the result a "milestone."

The timing of the announcement has raised significant concerns. Just one day prior, New York University professor Tristan Buckmaster published findings on a related problem with Levent Alpöge, a researcher at Anthropic. Buckmaster claimed that after contacting OpenAI to inquire about their progress, the conversation turned hostile. An OpenAI researcher reportedly asked, "Why would you ruin your career?" and responded with, "If you don’t want me to be nice, then I don’t have to be nice." OpenAI urged Buckmaster to credit only their model and drop Alpöge as a coauthor.

Buckmaster questioned whether OpenAI accessed his sessions on Codex, but the company grew evasive. OpenAI has denied accessing specific user data, stating they did not see "any of their work through any means until they released it publicly." However, they admitted that while unlikely, de-identified data from user interactions might have indirectly influenced their models.

The complexity of the situation is compounded by OpenAI’s own account of a hurried, expensive effort costing millions. The company stated they began working on the problem after hearing rumors on Twitter that other researchers were making progress on Millennium Prize problems. Researcher Sébastien Bubeck remarked, "We have such a strong model. Why don’t we try to solve also a Millennium Prize problem?"

Buckmaster disputed parts of the account, specifically that OpenAI researcher Bubeck asked him to remove Alpöge as a coauthor. OpenAI has largely ignored other allegations and directed inquiries to their blog post.

Regardless of the data access specifics, the rush to beat other researchers has shocked the mathematical community. Scooping is uncommon due to the specialized expertise required. Matthew Ballard noted that mathematics relies on an informal norm of trust, where researchers share incomplete ideas expecting them not to become competition. The fear that AI systems might steal ideas is described as "chilling."

Brendan Hassett from Brown University suggested that given the history of AI companies appropriating copyrighted work, researchers have a right to demand accountability regarding chat logs. Yang-Hui He expressed worry that mathematics under big companies is becoming too secretive, potentially reverting to a patronage-based era.

While most high-level problems won't receive such massive resources, Andras Juhasz of Oxford views this as a PR victory for OpenAI that questions the sustainability of AI's role in math. "Suddenly, 10,000 mathematicians jump on your problem," he said. Ultimately, OpenAI proved its models can compete at the frontier, but appears to have alienated the very community they hoped to impress.