AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
The Hard Line Response to Manipulation
"You win war, you win battles, but you never win wars," he says. "The opponent will just re jigger their example to satisfy whatever the compatibli responder is going to identify as the relevant difference." He argues that if an agent is manipulated ina rich sophisticated way, so that none of her internal architecture or anything, is any different than me sitting here in my room chatting with you right now,. assuming you think i'm a free and responsible agent... You might think i'm bananas, that other people have bt i believe it's true.