Evidence item

DeepMind's approach represents a paradigm shift in mathematical research. Instead of relying solely on traditional analytical methods, the team used physics-informed neural networks-a type of AI that incorporates physical laws directly into its learning process. This allowed the AI to discover new families of unstable singularities across three distinct fluid dynamics equations: the Incompressible Porous Media (IPM), Boussinesq, and Navier-Stokes equations.