Fields that I reference when thinking about AI takeover prevention — LessWrong