x
Video and transcript of talk on giving AIs safe motivations — LessWrong