This video explores how emerging deep learning frameworks like Hamiltonian and Lagrangian neural networks are revolutionizing the discovery of physical laws. Learn about a novel approach developed at MIT that uses automatic differentiation to compute high-order derivatives and integrates data from multiple dynamical systems to extract candidate equations of motion. Discover how this MASS framework not only formalizes classical mechanics but also reveals new symmetries and invariants that challenge traditional physics paradigms. The presentation is based on research by Xinghong Fu, Ziming Liu, and Max Tegmark from MIT's Department of Physics and Institute of Artificial Intelligence and Fundamental Interactions, titled "Do Two AI Scientists Agree?"