Open-Ended Innovation
Jeff discusses the potential of algorithms to mimic the sustained innovation seen in Darwinian evolution and human culture. He highlights a new approach to AI safety through Automatic Capability Discovery, which automates the exploration of a model's capabilities and failure modes. This method captures the collective insights of the scientific community, allowing for a more thorough understanding of AI systems and their unexpected behaviors.In this clip
From this podcast
Machine Learning Street Talk (MLST)
Jeff Clune - Agent AI Needs Darwin
Related Questions