I keep thinking about AI image models' effects on illustration and comparing/contrasting with things like GitHub's CoPilot for code. Inasmuch as someone starts

I keep thinking about AI image models' effects on illustration and comparing/contrasting with things like GitHub's CoPilot for code. Inasmuch as someone starts with an exact mental image and wants something matching that mental image, then no, it could be quite a while before an AI model is able to do that. Maybe six months, maybe six years, maybe never. As always with AI, it's hard to say. How often is it true that someone insists on something that exactly matches their mental image? I honestly don't know! If I write another novel, how picky am I going to be to be about the cover? If I get Stable Diffusion, or a future version of something similar, to generate a dozen images that match my text prompt, do I just pick the one I like best and lay some text over it? Or do I decide that no, I need something very specific, and hire someone to draw exactly what I want? I think one of the main criticisms of AI code generation is that you end up needing to be so specific with your "text prompts" that you're essentially writing the code yourself and letting the AI handle syntactic sugar. A sort of "AI IntelliSense™," if you will. But that's because we're developers. Like the illustrator in this piece, we have an exact idea of what we want, and nothing else will do. What if I were a non-technical person and could ask an AI model to give me code that collects names and email address into a database I can later use for a mailing list? Could an AI model do that? If not today, then surely soon. Isn't that a similar issue? Does it put me out of work? No, not immediately. But it does potentially eliminate an entire category of work, specifically the work that people generally use to learn how to be good developers. Will this cause problems with training future developers, since they have to rise above a certain skill level to exceed AI models? It might! I wonder if the same is true of illustrators. Maybe existing illustrators have their niche, but how hard will future illustrators have to work before they can exceed AI models?