Maia Fraser

I have a dual background in pure mathematics (Stanford PhD) and machine learning (Chicago PhD). Prior to starting my academic career at the University of Ottawa, I worked for several years in industry in high-performance computing at Supercomputing Systems AG. I'm the founder and director of Inter-Math-AI (IMA), an NSERC CREATE-funded graduate training program running since 2022 across four Canadian universities — focused on bringing social impact, ethics, and safety considerations into graduate training at the intersection of mathematics and machine learning. My machine learning research sits in ML theory, particularly translating principles across paradigms (e.g., semi-supervised learning to continual reinforcement learning) and applying learning-theoretic approaches to neuroscience and cybernetics. A complex adaptive systems perspective runs through this work, and has driven my focus over the last five years on AI safety — including serving as guest editor of the Bulletin of the AMS special issue "Will Machines Change Mathematics?", founding IMA, and helping to establish the Centre for Mathematical AI at the Fields Institute, where I sit on the steering committee. I see the math-ML intersection as particularly fertile ground for AI safety progress, and in particular feel it has a crucial role to play in mitigating existential risk from AI, both via technical safety and via principled policy and governance.

Projects

No grants recorded.