I don't know about the "top of my head" thing, I think that's kind of a Bob Ross sort of chaotic do instead of plan. Bob Ross I'm sure spent a lot of his life on trial and error, so he knows "if you mess this up you can fix it this way".
My process is as a designer for the past 20 years, which is why it's so similar to AI.
Someone gives me a "prompt": "I need a banner for the high school football team"
Immediately in my head I'm thinking "school colors are maroon, white, black, lets maybe do maroon background with an outward fade to black, find a vector art of a quarterback charging and match up to school colors, collegiate fonts, draw a quick football field, find or draw a football" and while I have aphantasia meaning I don't "see visions in my head", I know exactly how to lay it out because my mind thinks in words.
I've read Mastering Layout and such to help me design better. I learned all sorts of complex photoshop effects from hours of studying, watching tutorials and so on. As I improved I found I could one shot customer requests at a quick pace.
AI is taking a lot of that labor off my back and now I can produce better designs at a fraction of the time and my customers get a better product at the same price or even less if it was something more complex that I would've spent an hour or two on.
Y'know what's funny?
The longer I'm in this space, sitting in the middle of the AI art debate, the more I catch glimpses of a lot of peoples' processes when it comes to scratchin' out hand-drawn art.
As someone who also draws, I'm left with a handful of questions. A couple of observations. A few musings, if you will. Primarily:
When did it become so common to take mashups of reference images, cut and paste them together to make a new pose or composed image and trace your desired characters and such over it all? Hell, sometimes you just see them take screenshots or memes and just kinda scribble their stuff on top of it (reference the recent flux of Syndome memes).
Here I was the whole time over many years studying anatomy and shit, trying to draw everything off the "top of my head," like "real artists" supposedly do. Y'know, to demonstrate mastery of techniques and such. Meanwhile, we got all these "hand-made" advocates out here basically using serial killer collage notes to guide their drawing hands.
Did you get permission to use any of those references you chopped up? How is it functionally any different from what an AI learning model does? Is it simply that human hands touched a pencil or a stylus?
What changed? What makes one okay and not the other?
I'm just an old man with questions. Like what you like. Do whatever makes ya happy. I very much enjoy pretty imagery regardless if it came from a brush or a prompt, but I legitimately don't understand the difference here.