Everyone can generate a stunning image now. So the skill moved.

Midjourney (and the rest) made "make me something beautiful" a solved problem. Which means the value isn't in the generation anymore — it's in everything around it: → taste, to pick the 1 good frame out of 40 generic ones → art direction, so a batch feels like one brand instead of 40 random vibes → integration — actually getting those assets into a product, on-brand, at the right size and performant, not just sitting in a folder

I build with AI every day (I'm a developer, not just a prompter), and it's the same pattern everywhere: the model does the easy 80%, human judgment on the last 20% is the whole job. The people winning with AI aren't generating the most — they're the ones who know which output to throw away.

Generation is cheap now. Curation and integration are the craft.