This isn't the same thing. This is a commanded recital of a lack of capability, not that its confidence in it's answer is low. For a type of question the GPT _could_ answer, most of the time it _will_ answer, regardless of accuracy