Currently that's the only practical way to use it. It's very bad at doing specific things (as in narrowly defined stuff). It's MUCH better at giving.what you want if you can draw at least a clear but mediocre version of the final draft. All of them are incredibly awful at consistency so they aren't good for things like multiple angles of the same character or animation. Any attempt at that is a fever dream at best which is cool, but not really what you want in the vast majority of media.

Until these issues are solved it's mainly going to be a concept tool imo. I know a few professional and semi- professional artists who sketch a few things then generate like 20-30 different versions to help them concept out or send something back to the client so they can get clarification on what they want.

I used AI art recently as references for a commission I made. I think this will largely do what most tools have done. Raise the minimum bar of acceptable work this driving the quality up but the work amount will be the same overall, at least for the next few years.