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. Over the centuries, its borders have been expanded solely thanks to human ingenuity. At Google DeepMind, we believe that artificial intelligence can serve as a powerful tool for collaborating with mathematicians, enhancing creativity and accelerating discovery.
Today we are 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 inaugural partner institutions are:
- Imperial College London
- Institute for Advanced Studies
- Institute of Advanced Science Studies (IHES)
- Simons Institute for Computer Theory (UC Berkeley)
- Tata Institute of Fundamental Research (TIFR)
The initiative's partners will work toward shared goals of identifying the next generation of math problems ready for AI-driven insights, building the infrastructure and tools to support these advances, and ultimately accelerating the pace of discovery.
Google's support includes funding from Google.org and access to cutting-edge Google DeepMind technologies, such as an enhanced reasoning mode called Gemini Think Deeplyour algorithm discovery agent, AlphaEvolve, and our formal evidence completion system, AlphaProof. The initiative will create a powerful feedback loop between basic research and applied artificial intelligence, opening the door to deeper partnerships.
A pivotal moment for artificial intelligence and mathematics
The AI for Math initiative comes at a time of extraordinary progress in AI's reasoning capabilities; our work has progressed rapidly in recent months.
In 2024, our AlphaGeometry and AlphaProof systems achieved a silver medal at the International Mathematical Olympiad (IMO). Recently, our latest Gemini model, equipped with Deep Think, achieved a gold medal result at this year's IMO, successfully solving five out of six problems and scoring 35 points.
We saw further progress with another of our methods, AlphaEvolve, which was applied to over 50 open problems in calculus, geometry, combinatorics and number theory, with 20% of them improving previously best-known solutions. IN discovering mathematics and algorithmsinvented a new, more efficient method of matrix multiplication – a fundamental computation in computer science. For the specific 4×4 matrix multiplication problem, AlphaEvolve discovered an algorithm that used just 48 scalar multiplications, breaking a 50-year record set by Strassen's algorithm in 1969. information technologyhas helped researchers discover new mathematical structures that show that some complex problems are even more difficult for computers to solve than we previously knew. This gives us a clearer and more accurate understanding of the computational constraints, which will help guide future research.
This rapid progress demonstrates the rapidly expanding capabilities of artificial intelligence models. With this new initiative, we hope to explore how artificial intelligence can accelerate discovery in mathematical research and solve more difficult problems.
We are only at the beginning of understanding all that artificial intelligence can do and how it can help us think about science's deepest questions. We believe that by combining the deep intuition of the world's leading mathematicians with the innovative capabilities of artificial intelligence, you can open new research paths, deepening human knowledge and heading towards new breakthroughs in scientific disciplines.
