Getting decent output from AI is an art in ofitself. A lot of tinkering with parameters, getting close then trying again. Taking a good output, pulling it into an image editor to fix the broken bits, sometimes repeating that process multiple times until you get what you're looking for.

I think in the long run it'll be a tool that artists use, much like many digital artists today use tools that once were considered to be antithetical and unreconcilable with art