Paul Crowley comments on Strategic research on AI risk