Physics and the Puzzle of Artificial Intelligence with Zohar Nussinov

The 2024 Nobel Prize in Physics recognized John Hopfield and Geoffrey Hinton for breakthroughs in Machine Learning and Artificial Intelligence. This is a reminder that physics has long shaped the way we design and understand intelligent machines, and the mathematics underlying many modern approaches has been developed and honed in various areas of physics. The mathematical unity creates unexpected connections – for instance, the basic building block of a “neural net” has some links to quantum mechanics, while the optimization methods that power modern AI follow the same equations as the suspension system of your car! In this talk, we will explore how ideas from physics — from entropy and information to the behavior of complex systems like spin glasses — have influenced the rise of machine learning, and consider the open puzzle of why machine learning works so well.