An AI trained on snapshots of a complex physical system successfully discovered the underlying laws of physics without any help from humans.
April 23, 2026
Original Paper
Autonomous Emergence of Hamiltonian in Deep Generative Models
arXiv · 2604.20821
The Takeaway
Deep generative models can recover microscopic Hamiltonian parameters with 99.7% accuracy from raw data. The AI was not given any physical priors or formulas to start with, it simply observed the system in thermal equilibrium. This proves that machine learning can go beyond pattern matching to uncover the fundamental governing equations of nature. We can now use AI to discover the laws of physics for materials and systems we do not yet understand. It marks the beginning of AI as a primary discoverer of scientific laws rather than just a calculator.
From the abstract
The unprecedented predictive success of deep generative models in complex many-body systems, such as AlphaFold3, raises an epistemological question: do these networks merely memorize data distributions via high-dimensional interpolation, or do they autonomously deduce the underlying physical laws? To address this, we introduce a rigorous algebraic framework to extract the implicit physical interactions learned by generative models. By establishing an exact equivalence between the zero-noise limi