In response to critiques of Guaranteed Safe AI — LessWrong