Special guest session with Natasha Jaques (Assistant Professor, UW / Staff Research Scientist, Google DeepMind) a leading researcher in Social Reinforcement Learning and one of the pioneers of RLHF.

AI Safety is a Multi-Agent Problem: Natasha will discuss why current safety training paradigms fall short and make the case for a multi-agent framework as essential for provably safe LLMs. Session includes time for Q&A on her research, PhD applications, and careers spanning academia and industry.