Safety evaluations and standards for AI | Beth Barnes | EAG Bay Area 23
EAG Talks
00:00
Evaluating Models for Dangerous Things
The chapter discusses the importance of evaluating the safety of AI models to prevent existential risks. It proposes a plan to iterate through successive generations of systems to develop threat models and establish concrete safety mitigations. The chapter also explores the capabilities and potential threats posed by models, including self-sabotaging behaviors and the challenge of shutting them down.
Transcript
Play full episode
