This is non-sensical. You hold the paintbrush and move it across the medium with your own hand. It is entirely within the artists control. With a computer, you tell it what to do and within it's own constraints, be that p5.js or an llm whatever algorithm will produce some result.
When you use the pencil tool across the screen in photoshop, it works just as a pencil on paper.
The difference is not so clear as you imply.
When using a wacom tablet, using the pencil tool in photoshop is very similar to using a pencil. You're "instructing" the pencil tool to move by moving the pencil controller with respect to the tablet.
I know AI slop art when I see it, and I have no interest it making or "consuming" it. But I don't think I could construct a robust definition that would stand up to adversarial scrutiny.
Shelley Lake did not use p5.js nor an LLM for The Last Starfighter, and I didn’t use either of those things either. She may not have written the software she was using, and generally speaking I do write the software. I don’t know that it matters who writes the software, but it does make it easier to diffuse artistic credit if you use complex tools you don’t build yourself. I have done quite a bit of both physical painting and digital art and in both cases I’m deciding what gets done, designing the outcome, making the tools myself, and like a paintbrush, the computer doesn’t participate in any of that. The computer is just as much a tool as a brush, and is no more an artist than the brush is.
LLMs are brand new to art, they are not relevant to the top comment of this thread, and in terms of provenance of an image they are not understood by the public yet. I don’t feel that you’ve offered any clear thinking about how to attribute images produced using an LLM. It would be best to leave them out of the discussion for now, but FWIW they don’t change the calculus of computers vs intent and attribution much, they only muddy the water due to humans making all the training data.
I wrote the program that creates that. When I wrote it, my goal was to make a procedurally generated gas giant planet of some sort, but I didn't even know if what I was attempting would work (and for a period of about a month, it didn't work). I had/have only rudimentary control over the output. I can control the colors that go into it. I can control, to some degree, the scale of the "swirliness", but not the details of the swirliness. I can control how long I let the "swirliness" progress, and a few other things. But, I never know ahead of time how it's going to come out, and it's hard to escape the obvious fact that the computer has done nearly all of the work, and I only gave it the vaguest of directions in terms of aesthetics, although I did give it very specific instructions about how to do everything it did, those instructions were so complicated and so many, that I had next to no idea what specific output those instructions might produce, except in the most general terms. When I run the program, there's an anticipation ... "oooh, I hope this one comes out good!" but I don't know ahead of time whether it will or not.
You picked the subject matter and setup the camera and shading. You designed the software UI. You decided what level of control you wanted, and you can add more control if you want. I understand you’re using procedural rules and randomness, but that was your intention, and so the story that you don’t have control and don’t know what it’ll look like when run seems to me contrived at some level.
One of my main digital art projects centers on artificial evolution, which is notoriously difficult to control; I am deeply, intimately familiar with what you’re talking about, and I still say emphatically the computer is not the artist, I am, especially when I write the software. In a digital art project, the software is part of the art. You’re framing your control in terms of what happens only between starting up the software you wrote and getting one output image, but for some reason leaving out all the control and intention you exercised writing the software and deciding what kind of procedural pictures you wanted.
There’s a whole category of art called “generative art” that isn’t about computers specifically, but started long before computers existed, and is about setting up and releasing a system of rules, whether physical or logical. Generative art relies on some level of randomness, and the resulting artifact is rarely predictable. And yet, it is still deterministic. The rules might be complex, but the only intention comes from the humans involved who establish the rules, and then let them execute, and hold up the output as art.
Noticing the machine does the labor isn’t very relevant, IMO. Cars do a ton of complex work while driving with only vague directions, but we don’t anthropomorphize them, not seriously anyway. Knitting machines do all the complex work making clothes, but clothing designers don’t talk about collaboration with the machine, nor give design credit to the machine. Printers and copiers are very complex, but nobody waxes philosophical about them being collaborators. Just because an image might come out different every time a program runs that was designed to produce a different image every time it runs doesn’t make the machine an artist, it only shows that the computer is a useful, powerful, and reliable machine.
whether it’s a physical or digital medium the artist is expressing their creativity and manipulating the tools to realize that vision.