Automated assistants - History — LessWrong