Søren Elverlin comments on AI Safety: A Climb To Armageddon?