Geoffrey Miller comments on AI alignment researchers don’t (seem to) stack