I research how deep neural networks can learn continually, acquiring new skills without losing their ability to adapt. My work draws on concepts from statistical physics and dynamical systems theory to understand and improve the plasticity of learning systems.
I also study multi-agent reinforcement learning, focusing on the challenges of non-stationarity that emerge when intelligent agents learn and evolve together.
More broadly, I’m also interested in AI safety, climate change, effective altruism, protein folding, game theory, the evolution of cooperation, nuclear fusion, and AI for science in general.
Text box item sample content