Yudkowsky is too optimistic about how AI will treat humans. — LessWrong