Søren Elverlin comments on A long reply to Ben Garfinkel on Scrutinizing Classic AI Risk Arguments