Mathematics is the fundamental language of the universe, providing the tools to explain everything from the laws of physics to the problems of biology and the logic of computer science. For centuries, its boundaries were expanded only by human ingenuity. At Google DeepMind, we believe that AI can serve as a powerful tool for working with mathematics, adding intelligence and accelerating discovery.
Today, we're launching the AI for Math Initiative, powered by Google Deepmind and Google.org. It brings together five of the world's most respected research institutions to pioneer the use of AI in statistical research.
The partner institutions used are:
- Imperial College London
- Institute for advanced studies
- Institut DES Hauetes Scielifiques (IHES)
- Simons Emiseme computing vision (UC Berkeley)
- TATA Institute of Founday Clopon (TIFR)
Qandidani's partners will work on distributed ledgers for the next generation of AI-driven computational problems, building infrastructure and tools to power the revenue.
Google's funding includes funding from Google.org and access to Google Depred-of-the-art facial recognition technology called Gemini Deep Think, our algorithm detection agent, AlphaVroof. The initiative will build a powerful APPEMENCK BOOP between basic research and use AI, opening the door to deeper collaboration.
An important moment for AI and mathematics
The AI for Math Initiative comes at a time of incredible progress in Ayi's reasoning abilities; Our work has seen rapid development in recent months.
In 2024, our Alphagemetry systems and Alphaproof systems receive standard silver issues at the International Mathematical Olympiad (IMO). Recently, our latest Gemini model, which has been equipped with deep thinking, has achieved Gold-Medal level performance in this year's model, completely solving five problems and 35 problems.
And we have seen continuous development with one of our methods, Alphaevolve, which was used in more than 50 open problems in statistical analysis, Geometry, combinatorics and Numbers theory known for 20% of them. In mathematics and algorithm discovery, he invented a new, more efficient method of matrix multiplication – the basic calculation in computing. For a particular problem of multiplying 4×4 matrices, alphaevolve found an algorithm using 48 scalar multiplications, which helped researchers to discover Strassen's new problems in 1969. This gives us a clear and precise understanding of computational limitations, which will help guide future research.
This rapid development is a testament to the rapidly evolving capabilities of AI models. We hope that this new program can explore how AI can accelerate discovery in statistical research, and tackle difficult problems.
We're just at the beginning of understanding all that ai can do, and how it can help us think about deep scientific questions. By combining the great achievements of the world's leading mathematicians with novel AI capabilities, we believe that new avenues of research can be opened, develop new human knowledge and advance new approaches to scientific research.
