Mathematics serves as the foundational language of the universe, describing everything from the laws of physics to the logic of computer science. For centuries, human ingenuity alone has expanded its frontiers. At Google DeepMind, we believe AI can act as a powerful collaborator with mathematicians, enhancing creativity and accelerating discovery.
Today, we introduce the AI for Math Initiative, supported by Google DeepMind and Google.org. This groundbreaking effort unites five of the world’s most prestigious research institutions to pioneer the use of AI in mathematical research. The inaugural partner institutions include Imperial College London, Institute for Advanced Study, Institut des Hautes Études Scientifiques (IHES), Simons Institute for the Theory of Computing (UC Berkeley), and Tata Institute of Fundamental Research (TIFR).
Together, these partners will identify the next generation of mathematical problems ripe for AI-driven insights, build infrastructure and tools to power these advances, and ultimately accelerate the pace of discovery. Google’s support encompasses funding from Google.org and access to Google DeepMind’s cutting-edge technologies, including the enhanced reasoning mode Gemini Deep Think, the algorithm discovery agent AlphaEvolve, and the formal proof completion system AlphaProof. This initiative establishes a powerful feedback loop between fundamental research and applied AI, opening doors to deeper partnerships.
This comes at a pivotal moment for AI and mathematics, marked by remarkable progress in AI’s reasoning capabilities. In 2024, our AlphaGeometry and AlphaProof systems achieved a silver-medal standard at the International Mathematical Olympiad (IMO). More recently, our latest Gemini model with Deep Think reached gold-medal level performance at this year’s IMO, perfectly solving five of six problems and scoring 35 points.
We’ve witnessed further breakthroughs with AlphaEvolve, applied to over 50 open problems across mathematical analysis, geometry, combinatorics, and number theory. It improved previously best-known solutions in 20% of cases. In mathematics and algorithm discovery, it invented a more efficient method for matrix multiplication—a core computing calculation. For 4×4 matrices, AlphaEvolve discovered an algorithm using just 48 scalar multiplications, breaking the 50-year record set by Strassen’s algorithm in 1969. In computer science, it helped uncover new mathematical structures revealing that certain complex problems are harder for computers to solve than previously known, providing clearer understanding of computational limits to guide future research.
This rapid progress showcases the fast-evolving capabilities of AI models. We hope this initiative will explore how AI can accelerate discovery in mathematical research and tackle even harder challenges. We’re merely at the beginning of understanding AI’s full potential and how it can help us address science’s deepest questions. By combining world-leading mathematicians’ profound intuition with AI’s novel capabilities, we believe new research pathways will emerge, advancing human knowledge and driving breakthroughs across scientific disciplines.