x
Theory of (advanced) agents — LessWrong