Craig Mattson comments on Is AI risk assessment too anthropocentric?