Co-founder Kasra Rafi shares the vision behind our newest collaborative research initiative.
Launched last spring under the leadership of former Deputy Director, Kasra Rafi, the Fields Centre for Mathematical AI was created to enhance the link between mathematics and artificial intelligence. Housed at the Fields Institute, the Centre brings together mathematicians, computer scientists and machine learning researchers from across disciplines to investigate two complementary questions: how mathematics can make AI systems safer and more transparent, and how AI can become a genuine tool for advancing mathematical discovery.
We’ve set out to create a collaborative environment where mathematicians and AI researchers can learn from each other and tackle problems together. The Centre aims to bridge the gap between pure mathematics and artificial intelligence so that each field can amplify the other. We want to see mathematicians contributing to the development of AI, for example by bringing rigorous mathematical formalism to improve AI trustworthiness and alignment, and we also want to see how advanced AI tools can accelerate discovery in pure mathematics. Ultimately, we see the Centre as a hub where new ideas emerge from the synergy of math and AI, leading to breakthroughs that neither field could achieve alone.
By focusing on building a strong community and providing the right platforms. This means organizing interdisciplinary workshops, seminars and collaborative research programs that bring mathematicians and AI scientists into the same room. By hosting regular events (for example our Mathematical AI seminar series) and working groups, we create opportunities for people to share knowledge and start new projects together. We are also partnering with institutes like the Vector Institute and with institutions around the world to connect with a broader AI community. The idea is to provide a supportive infrastructure, from computational resources to expert networks, so that ambitious joint math-AI projects can take off.
Toronto is uniquely suited for this initiative. It’s home to world-leading AI research at the University of Toronto, the Vector Institute and a thriving network of start-ups and labs, alongside top universities with strong math departments. This means we have a critical mass of expertise right on our doorstep, from AI scientists to pure mathematicians, all within the same city and province.
Fields itself provides the ideal home base: it has decades of experience connecting research communities, hosting collaborative programs and supporting interdisciplinary networks. Its infrastructure and reputation in the mathematical sciences make it a natural setting for a centre that spans disciplines. Finally, Fields is centrally located in Toronto and has connections across Canada and internationally. This helps us draw in talent and participants not just locally but from around the world.
AI runs on mathematics. The qualities people care about—fairness, reliability, interpretability—only become clear once they’re expressed mathematically. Mathematics lets us move from “it works” to “we understand why it works.” Future progress in AI will depend not just on more data, but on new mathematical insight.
These concerns are exactly why academics should be involved in building these tools, not standing outside and rejecting them. If mathematicians and educators help design AI, we can steer it toward systems that support learning, find the right references and help us check correctness, rather than tools that only compete with students or faculty. The best way to address the concerns is to help build AI that serves academic values, not to opt out.
A good outcome is that, after a few years, the Centre is a stable community where mathematicians and AI researchers work together and we can point to concrete results that started here, papers on the foundations of AI, math problems solved with AI assistance or formalization tools people actually use.
It would also be a success if we train a cohort of young researchers who are genuinely bilingual in math and AI, and if people start to think that if you have a math and AI idea, you take it to Fields. That would mean we have created both new knowledge and a durable culture around it.
The Centre for Mathematical AI is new and there is a lot of optimism and energy around it. We are in the early days which means we are learning and adapting, but it is also the right time for people to get involved. This is not an exclusive group. It is open to anyone curious about the interplay of math and AI, whether you are a student who wants to learn, a researcher with an idea or an industry practitioner looking to collaborate.
