BYLINE: Perimeter Institute for Theoretical Physics
Newswise — Perimeter Institute for Theoretical Physics is pleased to announce the launch of the Physics-Driven AI Initiative (PhysDAI), a strategic vehicle for Perimeter’s AI-enhanced research programs.
“AI is already changing the way mathematics is being done. In the next two to five years, it will change the way we do research in physics drastically and may lead to new breakthroughs. Perimeter wants to be at the forefront of that — of how AI will change physics research in a radical way,” said Perimeter Institute Executive Director Marcela Carena.
Perimeter’s approach to AI takes into account a wide swath of opportunities and risks related to the burgeoning new technology. AI tools have proven very capable at certain scientific tasks, including processing enormous datasets produced by observatories and experiments, and carrying out complex mathematical calculations. At the same time, scientific users of AI are grappling with challenges like interpretability (the degree to which we can understand how AI reaches its answers) and the need to verify its results. Physics expertise can lend a hand in both these areas.
“With AI, the coding bottleneck that once slowed theoretical work is largely gone, and we now have tools that can connect ideas across fields in ways no single researcher could. One of Perimeter’s great strengths has always been the interplay among its nine research areas, and AI is turbocharging that process, bringing discoveries we once imagined for the distant future within much closer reach. It’s an incredibly exciting time to be a researcher,” said Research Faculty member Freddy Cachazo, who sits on the PhysDAI steering committee.
PhysDAI at Perimeter has three initial pillars of operation, all of which put human scientists at the centre of the AI strategy.
The first pillar provides dedicated funding to hire postdoctoral researchers working at the intersection of physics and AI. The initial PhysDAI cohort of postdoctoral researchers arrived in Fall 2026, several of whom hold joint appointments with partner institutions, including the University of Toronto and the NSF Institute for Artificial Intelligence and Fundamental Interactions (IAIFI).
The second pillar brings long-term visiting researchers with expertise in AI to Perimeter’s unique research environment. It also funds seminars and workshops open to all of Perimeter’s researchers seeking to build up their research capacity using AI.
The final pillar supports the creation of new permanent faculty positions dedicated to the field, some of which will be in partnership with collaborating Canadian universities. Perimeter expects the first of these new faculty members to arrive in the 2026-27 academic year.
With PhysDAI, Perimeter is playing to its strengths, bringing in experts not solely trained within the AI commercial ecosystem, but rather those with a background in theoretical physics who are now engaging with AI on their own terms.
“It’s about finding experts that have physics in their DNA, but who have now put enormous efforts into AI. Some of our own people are going through that metamorphosis right now, making the transition to AI rapidly. But we are also bringing in people who have already gone through this transition, or who are at different stages of building a connection with AI,” said Carena.
The goals of the initiative are twofold and can be summarized as AI for physics and physics for AI. The first applies AI tools to existing problems in theoretical physics to accelerate breakthroughs. The second applies lessons from physics to inform our understanding of how AI tools work and improve the next generation of AI models.
“We are embarking completely on the AI train. We are lucky to have already done so,” said Carena. “When I talked about this a year ago, I was saying the train is about to leave the station, and we’d better jump on. The train has left the station now for sure. But we are on it, and the point is that we have to keep being the sharpest we can.”
AI is a national priority for Canada, as it is for countries around the world grappling with the implications of this new technology. Perimeter Institute aims to contribute to the national and international conversation by bringing its own unique research-grounded insights to the table.
“I see Perimeter as leading theoretical research – in this case, theoretical research in AI – at an international level, and therefore delivering a significant contribution to Canada, and helping elevate Canada. Of course, there are many people doing AI research in Canada, but we are doing it from our own foundational research-focused niche, which is unique,” said Carena.
To learn more about PhysDAI at Perimeter, visit the initiative homepage here.
