x
Alignment to Continuation: Preserving Human Playability Instead of Values — LessWrong