> I thought was fully obvious from the context I gave it.

Lot of people think they have given the right instructions but in most cases people miss some crucial points and that leads the model in the wrong direction, then the same people complain AI is not good.