The Simons Foundation has launched the Simons Collaboration on the Physics of Learning and Neural Computation, aimed at unraveling the complexities of artificial intelligence (AI) and large neural networks. This multidisciplinary effort will harness physics, mathematics, computer science, and neuroscience to investigate how these systems learn, reason, and generate insights. Surya Ganguli, the collaboration’s director from Stanford University, emphasizes the need to decode the “black box” of AI, which, despite its rapid advancement, remains poorly understood.
With funding of up to $2 million annually for four years, the collaboration will explore the fundamental principles underlying AI by treating it as a complex physical system. This initiative seeks to connect researchers from top institutions, enhancing our understanding of data structures, learning dynamics, and neural architectures. Notable members include leading scientists from esteemed universities, all committed to advancing the knowledge of AI’s foundational mechanisms.
Source link
