Why mathematicians are worried about AI
What's the story
A group of mathematicians has warned about the growing influence of technology companies in their field, specifically through artificial intelligence (AI). The warning comes in the form of a declaration titled "Leiden Declaration on Artificial Intelligence and Mathematics." It was created by 16 researchers over eight months after a conference at Leiden University in September 2025. The International Mathematical Union has endorsed this declaration.
Concerns raised
AI threatening the 'characteristic values' of mathematical research
The Leiden Declaration, which has already garnered hundreds of signatures, warns that recent advancements in AI are threatening the "characteristic values" of mathematical research. These threats often disproportionately affect students and early-career mathematicians, thereby jeopardizing the long-term future of the discipline. The declaration also highlights how AI models can generate plausible but unreliable arguments that are hard to distinguish from correct mathematical proofs.
Literature concerns
Risk of cluttering literature with incorrect results
Leslie Ann Goldberg, head of computer science at the University of Oxford, warned about the risk of cluttering literature with incorrect results due to cheap production of inaccurate AI-generated drafts. She said these errors could propagate as new results are built on faulty foundations. The declaration also highlights that models trained on published works often return outputs that don't properly cite human contributions, further threatening research integrity.
Disruption risks
Declaration highlights potential disruptions in hiring and funding processes
The Leiden Declaration warns that the use of AI could disrupt traditional mechanisms for hiring, funding, and recognition in mathematics. This is particularly concerning for researchers who don't have access to or are unwilling to use technologies controlled by organizations whose values they don't share. The declaration also criticizes informal communication channels like press releases or blog posts for mathematical research as they can lead to oversimplification and misrepresentation of AI tools' significance.
Autonomy concerns
Concerns over tech companies' growing influence in mathematics
The declaration also expresses concern over the increasing involvement of technology companies in mathematical research, which could threaten its autonomy. This is especially true as university budgets come under pressure and researchers may feel more inclined to collaborate with tech companies on asymmetric terms. The authors of the declaration have also called for professional mathematical organizations to establish guidelines for using AI and other automated tools in publication and review processes.
Guidelines suggested
Recommendations for mathematicians and professional organizations
The authors of the declaration have recommended that mathematicians transparently disclose their use of AI tools and take responsibility for the correctness of their work. They should continue crediting human authors while properly attributing work even if AI tools make that difficult. Professional mathematical organizations can support peer-reviewed publications and protect researchers' author rights by utilizing licensing agreements that prevent their work from being used as AI training data without consent.