Scared about the future of AI — LessWrong