OpenAI's Academic Showdown: Feud With Mathematicians Escalates

A collective of twenty-five leading mathematicians, all recipients of the prestigious Fields Medal, has issued an open letter articulating profound concerns that artificial intelligence laboratories are jeopardizing the integrity of their intellectual work. This alarm stems from an observed competitive drive among AI labs, seemingly racing to develop solutions to celebrated mathematical problems.

Specific incidents underscore these growing apprehensions. Recently, NYU professor Tristan Buckmaster publicly accused OpenAI of pressuring him to withhold credit from a collaborator working for Anthropic, who had contributed to solving a significant mathematical problem. Buckmaster questioned whether OpenAI leveraged their prior work with its AI model, Codex, to generate a groundbreaking proof during an intense period of inference. Furthermore, OpenAI subsequently withdrew its sponsorship from a math event at CalTech following criticism from researchers at the university, indicating broader tensions within the academic community.

While the prospect of AI models solving complex mathematical challenges holds immense potential for humanity, the signatories of the letter argue that this benefit can only be realized if such solutions are comprehensible and communicable within the mathematical community and, by extension, to the wider world. They criticize the current trend where “Often these 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 of others.” The unverified status of OpenAI’s controversial proof exemplifies this concern. This practice, they warn, gives rise to severe questions of attribution and plagiarism, akin to challenges in other creative professions. Moreover, without the dedicated mathematicians to develop and integrate AI-conceived ideas into the mathematical canon, the crucial human transmission chain of mathematical knowledge would be irrevocably broken, leaving these ideas incomplete.

A palpable fear among mathematicians is that the culture of open research will be undermined. Many are becoming paranoid, wondering if their own use of Codex was unknowingly fed into OpenAI’s newer models. The current dynamic incentivizes secrecy, as frontier labs possess the financial means to deploy large language models (LLMs) to beat original researchers to a proof, potentially costing tens of millions of dollars. This situation directly threatens the collaborative and open spirit that has historically driven mathematical advancement.

This letter builds upon the foundation laid by the Leiden Declaration, a document released by a working group of mathematicians in June, which also examined how LLM-generated proofs would transform their field and offered a set of recommendations for mathematicians, institutions, and policymakers alike.

As AI tools increasingly reshape workflows in fields like software engineering, mathematicians are finding renewed justification in what they term the “work around the work.” This perspective emphasizes that the true value in mathematics transcends mere proofs and individual credit. Instead, it encompasses the intricate intellectual superstructure that fosters students, inspires new questions and ideas, and meticulously integrates these into the broader fabric of human civilization.

The mathematicians deliver a stark warning to those outside their discipline: “The issues the mathematical community faces now are similar to issues that other scientific and creative professions are facing, and indicate issues that all of humanity might face: how to make sure that, as AI changes the way work is done, we do not lose sight of what that work was meant to achieve in the first place.” They assert that if one disregards the cutthroat realm of high-stakes mathematical proofs, it is only a matter of time before similar challenges confront their own fields of interest.