Mathematics is the fundamental language of the universe, providing the tools to describe everything from the laws of physics to the intricacies of biology and the logic of computer science. For centuries its limits have been extended by human ingenuity alone. At Google DeepMind, we believe that artificial intelligence can act as a powerful tool for collaborating with mathematicians, increasing creativity and accelerating discovery.
Today we’re introducing the AI ​​for Math Initiative, powered by Google DeepMind and Google.org. It brings together five of the world’s most prestigious research institutions to pioneer the use of artificial intelligence in mathematical research.
The founding partner institutions are:
- Imperial College London
- Department of Advanced Studies
- Institut des Hautes Études Scientifiques (IHES)
- Simons Institute for Theory of Computing (UC Berkeley)
- Tata Institute of Fundamental Research (TIFR)
The initiative’s partners will work toward the shared goals of identifying the next generation of mathematical problems ripe for AI-powered insight, building the infrastructure and tools to drive these advances, and ultimately accelerating the pace of discovery.
Google’s support includes funding from Google.org and access to Google DeepMind’s advanced technologies, such as an enhanced reasoning mode called Gemini Deep Think, our algorithm discovery agent, AlphaEvolve, and our formal proof completion system, AlphaProof. The initiative will create a powerful feedback loop between basic research and applied AI, opening the door to deeper partnerships.
A defining moment for artificial intelligence and mathematics
The AI ​​for Math Initiative comes at a time of remarkable advances in AI reasoning. our own work has seen rapid progress in recent months.
In 2024, our AlphaGeometry and AlphaProof systems achieved a silver medal standard at the International Mathematical Olympiad (IMO). Recently, our latest Gemini model equipped with Deep Think achieved a gold medal level performance at this year’s IMO, solving five of the six problems perfectly and scoring 35 points.
And we’ve seen further progress with another of our methods, AlphaEvolve, which was applied to over 50 open-ended problems in mathematical analysis, geometry, combinatorics, and number theory, improving on the previous best-known solutions in 20% of them. In mathematics and algorithm discovery, it has invented a new, more efficient method of matrix multiplication—a core computation in computing. For the specific problem of multiplying 4×4 matrices, AlphaEvolve discovered an algorithm that used only 48 scalar multiplications, breaking the 50-year-old record set by Strassen’s algorithm in 1969. In computer science, it helped researchers discover new mathematical structures that show certain complex problems previously known to be even more difficult for computers. This gives us a clearer and more precise understanding of computational limits that will help guide future research.
This rapid progress is a testament to the rapid development capabilities of AI models. We hope that this new initiative can explore how artificial intelligence can accelerate discovery in mathematical research and tackle more difficult problems.
We are only at the beginning of understanding all that AI can do and how it can help us think about the deepest questions in science. By combining the deep intuition of world-leading mathematicians with the new capabilities of AI, we believe that new avenues of research can be opened, advancing human knowledge and moving towards new breakthroughs across scientific disciplines.