My thoughts on the social response to AI risk — LessWrong