Neural Networks for Mathematical Discovery

Explore how artificial intelligence is revolutionizing mathematical research in this 58-minute Presidential Lecture that surveys cutting-edge applications of neural networks in discovering mathematical examples and counterexamples. Learn about the surprising mathematical objects that have historically captivated researchers, from the icosahedron and aperiodic tilings to the Mandelbrot set, and understand how these discoveries have spawned entirely new mathematical fields. Discover recent breakthroughs where neural networks assist mathematicians in their search for interesting mathematical phenomena, with insights into why this represents only the beginning of a new era in AI-assisted experimental mathematics. Gain perspective on the future of mathematical discovery as artificial intelligence tools become increasingly sophisticated in identifying patterns and generating hypotheses that human mathematicians might overlook, potentially leading to a flourishing of computational approaches to mathematical exploration in the coming years.