Comment by andsoitis

> get this question correct

I am willing to bet a million dollars that it is unlikely any single model will ever be able to answer any question correctly.

The implications then are that one cannot use a single question evaluate whether a model is useful or not.