A curated, occasional list of essays, articles, lectures, and podcasts on
the relationship between mathematics and AI. The selection spans a wide
range of positions, from strong optimism to serious reservations, and many
that resist easy categorization. Restricted to sources that are free to
read or watch. Sorted newest first within each section.
These are personal picks. Inclusion does not imply endorsement, and does
not reflect the views of PIMS or UBC.
Sep 2026ICIAM Statement on Mathematics and Artificial Intelligence. International Council for Industrial and Applied Mathematics. Argues for active engagement rather than resistance or uncritical acceptance, and for mathematics as the source of verification, error control and uncertainty quantification in AI systems. An extended fourteen-page version is linked from the announcement.
Sep 2026Headlines and inside stories: understanding and trust in AI for mathematics, science, and engineering. Tapio Schneider (Caltech), guest post on Terence Tao's What's new. Argues that AI-generated proofs should advance conceptual understanding rather than only settle truth values, and extends the argument to science and engineering, where empirical verification is delayed. Clarifies that the announced Navier-Stokes result concerns a forced variant.
Sep 2026Advisory Group on Mathematics and Artificial Intelligence. Independent group hosted at the Institute for Advanced Study: François Charles, Camillo De Lellis, Timothy Gowers, Martin Hairer, Nikhil Srivastava, Ulrike Tillmann, Ravi Vakil, Edward Witten and Melanie Matchett Wood. Formed to advise AI companies on their interactions with mathematical research, and currently advising OpenAI on coordinating the release of results from an internal model.
Sep 2026A CERN for AI-assisted science? Dimitris Koukoulopoulos (Université de Montréal), guest post on Terence Tao's What's new. Argues for publicly funded frontier AI infrastructure for academic research, citing Canada's sovereign compute commitment and the EU AI Gigafactories.
Sep 2026A Severe Misalignment of AI in Mathematics. Declaration signed by 25 Fields Medallists, including Terence Tao, Peter Scholze, Maryna Viazovska, June Huh, Martin Hairer, Maxim Kontsevich, James Maynard and Manjul Bhargava. Argues that the use of mathematical problems as AI benchmarks is misaligned with the needs of mathematics. See the responses listed above.
Jul 2026Thoughts about the Leiden Declaration. Timothy Gowers (Collège de France / Cambridge), Gowers's Weblog. A response to the Declaration listed below.
Mar 2026Shaping the Future of Mathematics in the Age of AI. Johan Commelin, Mateja Jamnik, Rodrigo Ochigame, Lenny Taelman, Akshay Venkatesh, arXiv:2603.24914. Recommendations across values, practice, teaching, technology, and ethics; companion to the Leiden Declaration.
Sep 2026AI Has Solved One of Math's $1 Million Millennium Prize Problems. Quanta Magazine. OpenAI announced an AI-found finite-time singularity for the forced 3D Navier-Stokes equations, twelve hours after a team of Tristan Buckmaster and Levent Alpöge announced results on closely related problems. Both built on a strategy developed by Diego Córdoba and Luis Martínez-Zoroa. The unforced Millennium Prize problem remains open. Priority and credit are disputed.
Sep 2026Formalizing Fermat's Last Theorem. Anthropic, research blog. Reports a complete Lean formalization of the Wiles proof produced largely autonomously by Claude over 11 days, running to roughly 13 million lines and 29,500 intermediate theorems. Autoformalization of an existing proof rather than a new result. See the response below.
Sep 2026FLT: Anthropic has beaten me to it. Kevin Buzzard (Imperial College London), Xena Project. Independent verification and mathematical assessment from the leader of the FLT formalization project.