Sufficiently advanced Artificial Intelligence - History — LessWrong