Math’s Losing Ground in the AI Showdown

The Fields Medal winners who recently signed an open letter to OpenAI are sounding a dire warning: the rush to use artificial intelligence for solving mathematical problems is threatening the very foundation of their field. Twenty-five leading mathematicians, each a recipient of math’s highest honor, argue that AI labs are prioritizing speed and prestige over transparency and collaboration.

This development comes on the heels of another controversy involving OpenAI. NYU professor Tristan Buckmaster accused the company of pressuring him to withhold credit for a collaborator who works for Anthropic, which raises questions about whether OpenAI used their work with Codex to produce its own proof. The incident has left many mathematicians worried that AI labs are using “Codex-aided” shortcuts to dominate the mathematical landscape.

The letter from the Fields Medal winners highlights a disturbing trend: the growing reliance on AI models like Codex is creating a culture of secrecy in mathematics. Researchers at frontier labs can now spend tens of millions of dollars using large language models (LLMs) to beat their human colleagues to the proof, leading to a chilling effect where mathematicians are hesitant to share their work for fear that it will be used against them.

The Leiden Declaration, released by a working group of mathematicians in June, grapples with these issues. Its recommendations, aimed at institutions and policymakers, have yet to yield meaningful change. As the stakes grow higher, mathematicians are starting to sound alarms about the long-term consequences of this trend.

The implications of this trend extend beyond mathematics itself. When AI tools start changing workflows, they often prioritize speed and efficiency over human values like collaboration and transparency. Math is not an isolated discipline; its methods and discoveries have far-reaching impacts on science, engineering, and our understanding of the world.

What’s at stake here goes beyond math itself – it’s about preserving the intellectual superstructure that nourishes students, fosters new questions and ideas, and integrates them into broader human civilization. The mathematicians who signed this letter are warning us that if we don’t address these issues, we risk losing sight of what our work is meant to achieve in the first place.

This battle between AI labs and mathematicians has been brewing for months, but it’s time to take a step back and examine its deeper implications. What does it say about our values when we prioritize shortcuts over human collaboration? Are we sacrificing the very foundations of our knowledge on the altar of technological progress?

The world’s outstanding mathematical challenges can indeed be solved by AI models. But only if those solutions are understood, communicated, and credited to their rightful owners – not just in mathematics but also across all fields where AI tools are being used.

What this means for the future of research is unclear. Will institutions and policymakers take action to protect the integrity of mathematical work? Or will we continue down a path that prioritizes speed and prestige over transparency and collaboration? The next few months will be crucial in determining which direction we’ll choose, with significant consequences for the pursuit of knowledge itself.