My mental model has been increasingly one of mining minerals or gems. Prompting and iterating is like slowly chipping away the rock of a deep cave with a pickaxe. You then find a gem after many different prompts. Eventually you exhaust that particular area that you're trying to prompt.

The best quality things are incidental to how the model was trained and the training data, and it is not consistent in a combinatorial way as one might expect. Still, there are a lot of interesting things that are going to be found and are going to go viral, as people find interesting gems.