Verifier Challenges
Eliezer discusses the complexities of creating a reliable Verifier for AI systems, emphasizing that the difficulty increases with the intelligence and alien nature of the AI. He highlights that if it's hard for humans to discern truth, training AI to be accurate becomes even more problematic. The conversation delves into the nuances of trust and manipulation in interactions between humans and advanced AI.In this clip
From this podcast
Lex Fridman Podcast
Eliezer Yudkowsky: Dangers of AI and the End of Human Civilization | Lex Fridman Podcast #368
Related Questions