NYU to Participate in Federal Effort to Build AI Collaborator for Mathematics
A major DARPA award puts NYU at the forefront of one of the most ambitious bets in science: that AI can help mathematicians crack problems that have persisted for generations
In 2024, the Nobel Prize in Chemistry recognized breakthroughs in AI-driven protein folding — how proteins take the shapes that determine their function in the body — in a discovery that transformed biology.
Researchers now believe artificial intelligence could do something similar for mathematics, the discipline that underpins modern science, engineering, and computing. New York University is participating in that effort.
NYU has been selected as one of 13 teams in the inaugural cohort of the Exponentiating Mathematics (expMath) program, a newly launched initiative from the Defense Advanced Research Projects Agency (DARPA), the U.S. Department of Defense’s independent research arm known for funding high-risk, high-reward technologies.
The NYU team's project is led by Principal Investigator (PI) Claudio Silva, Institute Professor at NYU Tandon School of Engineering and a professor in the NYU Center for Data Science, and Co-PI Chinmay Hegde, an associate professor at NYU Tandon.
“Our project aims to build AI assistants that can work alongside mathematicians, translating informal proofs into formally verified mathematics and helping break complex problems into manageable steps,” said Silva, a pioneer in data visualization and computational science. “AI systems have historically struggled with the rigorous logical reasoning that mathematics requires. By combining machine learning with formal verification, we hope to make AI a useful collaborator in mathematical discovery. ”
The selection underscores NYU’s leadership at the intersection of artificial intelligence and mathematics as the university expands its investment in the mathematical and computational sciences.
In November 2025, NYU announced the creation of the Courant Institute School of Mathematics, Computing, and Data Science, uniting the Department of Mathematics, the Center for Data Science, and an expanded Department of Computer Science.
“This award is a testament to the extraordinary talent we have built at NYU across artificial intelligence, mathematics, and engineering,” said Juan de Pablo, NYU’s Anne and Joel Ehrenkranz Executive Vice President for Global Science & Technology and Executive Dean of NYU Tandon. “What Claudio, Chinmay and their colleagues are undertaking is exactly the kind of bold, cross-disciplinary work that defines NYU’s current scientific research agenda, and that we have been building towards through the newly-united departments of data science, computer science, and engineering since I joined in Fall of 2024.”
The stakes are enormous. Several of the Millennium Prize Problems — including the Riemann Hypothesis, the Navier–Stokes equations, and the Birch and Swinnerton-Dyer Conjecture — have resisted generations of mathematicians.
The NYU-led team intends to develop an automated pipeline to translate between LaTeX, the typesetting system mathematicians use to write papers and proofs, and Lean, a proof-assistant programming language used to formally verify mathematical arguments. Today, this translation is largely done by hand and can take experts anywhere from a day to multiple weeks to formalize a single page of mathematics.
The project’s AI assistants will be designed to learn how individual mathematicians write and reason about proofs, enabling more natural collaboration in the research process.
“Eighteen months ago, the idea that AI could operate at the level of research mathematics would have seemed like science fiction,” said Hegde, whose research focuses on machine learning for scientific discovery. “The pace of change has been extraordinary, and we’re still at the very beginning.”
In addition to Silva and Hegde, the team includes NYU Tandon’s Pavel Izmailov; NYU’s He He, and Sam Westrick; and Eduardo Teixeira, the Grayce B. Kerr Professor of Mathematics at Oklahoma State University and recipient of the ICTP-IMU Ramanujan Prize.
The team spans expertise in AI, human-computer interaction (HCI), formal methods, and mathematics. He He, NYU Associate Professor of Computer Science and Data Science, co-authored work on AlphaGeometry — an AI system capable of solving Olympiad-level geometry problems — that was published in Nature, an early demonstration of AI tackling advanced mathematical reasoning.