It was trained on human data about the same subject so we have every reason to expect it'd complain