Eigenhall·Eigenspace, 1st floor
The Rapid advance of AI in Math and Physics: Implications for Recursive Self-Improvement
Theoretical physicist, professor, and serial founder (Genomic Prediction, SuperFocus.ai); writes Information Processing and hosts the Manifold podcast.
40.0/40 attending (+27 waitlisted)
Chance I’d attend
Frontier LLMs are increasingly useful to researchers in math and theoretical physics. In some cases, important open problems have been solved by AIs. I review these developments and discuss their relevance to model Recursive Self-Improvement (RSI). Reasoning about model architectures and training methods strongly resembles reasoning in math/physics research. Hence, AIs may soon propose, code, and test improvements to their own architecture and training.
30 minute talk followed by questions