Mathematicians warn of AI threats to profession as industry encroaches

A group of mathematicians has issued a declaration warning about the growing influence of the tech industry and artificial intelligence on their field. This declaration, published on June 2, 2026, by 16 researchers after an eight-month effort, highlights the challenges AI poses to mathematics research. The International Mathematical Union has endorsed this "Leiden Declaration on Artificial Intelligence and Mathematics."

The declaration points out several key concerns. Firstly, AI models can produce plausible but unreliable mathematical arguments, making it difficult for reviewers to distinguish correct proofs from incorrect ones. This jeopardizes traditional standards of correctness and verifiability. Secondly, AI models often fail to properly cite human works they synthesize and are trained on data obtained through questionable means, including copyright violations.

Thirdly, the declaration warns that the use of AI may be incentivized for its own sake, disrupting hiring, funding, and recognition mechanisms in mathematics. It also raises concerns about the increasing involvement of technology companies in mathematical research, which could threaten the autonomy of the discipline, especially as universities face budget pressures.

Fourthly, the declaration criticizes the communication of mathematics research through informal channels like press releases, often without proper scientific evaluation. This can lead to oversimplification and exaggerated claims about AI tools, overshadowing human contributions.

Mathematicians involved, such as Kevin Buzzard of Imperial College London, emphasize the need for a thoughtful response to AI's disruption of their field. Michael Harris of Columbia University, an author of the declaration, notes that the tech industry operates on commercial logic, which often conflicts with the values of mathematics. He believes the declaration aims to reclaim the narrative about the values and goals of mathematics from the AI industry.

While acknowledging the "remarkable" achievements of AI, such as OpenAI's recent claim of disproving a mathematical conjecture, the declaration authors and other mathematicians raise concerns about transparency. Rodrigo Ochigame of Leiden University, another author, points out that OpenAI did not disclose crucial information about prompts, training data, or computational resources used. He highlights that proprietary AI models limit scientific assessment, as "basic information needed to assess the scientific meaning of the result is kept secret."