I’m not as anti-AI as a lot of my cohort (lefty artists). I’ve never been morally against art made by remixing other artists’ work, so it’s hard for me to be angry at the “stealing” AI does to make images. If AI images were only a toy, something people play with but don’t sell, I wouldn’t be bothered by AI images remixing other images at all.
But it’s not just a toy, and I am worried by the economics of it.
AI image generation makes images close to instantly, but it makes those images by remixing work by human artists who took hours, days, weeks to make those images. The styles AI imagery apes might have taken someone decades to develop.
Those instantly-generated AI-generated images then compete in a marketplace against human artists.
That’s not sustainable for the humans.
I have no problem at all with human artists training themselves on art other humans have made. In comics, that’s a primary way most of us learn.
But after I’ve learned from other cartoonists, I can’t instantly produce images. So when I compete with them, it’s on even ground.
When artists are competing with other artists, that can be tough, but the result isn’t that there’s less work for artists overall.
But the more AI enters the art market, the less work is left for human artists. AI simultaneously depends on our work existing to be remixed, and makes paying us for our work obsolete.
Second thought: I personally don’t feel threatened by AI (although I’m some combination of amused by and tired of people mistaking my work for AI).
But I really worry about the younger generation of illustrators. I think a lot of “entrance level” illustration work that used to exist is increasingly being done with AI.
AI is stealing.
And it’s definitely not AI.
“But I really worry about the younger generation of illustrators. I think a lot of “entrance level” illustration work that used to exist is increasingly being done with AI.”
That is true in way too many fields, though illustration is a good example. My daughter got a degree in animation from a top tier school, just in time for AI to destroy entry level positions there. Since AI can produce work at about 70% of the quality for 10% of the cost, so guess what most studios are doing.
I…. don’t think AI created images are a problem. Which I understand is a very unpopular take among creatives, but I think the hate is very myopic.
This conversation is a variation on a theme, from how the printing press put scribes out of work (but mainlined books) to “canned music” replacing orchestral pits at theatres (but mainlined movies) to Digital Art replacing Physical Art (but mainlining custom art) to now, when AI is replacing digital artists (but allowing anyone to make good art).
It pays to remember that the concept of a digital artist is younger than I am. The idea that society will not function without people being able to charge hundreds of dollars for commissions that it took them hours to put together is belied by the millennia we had without them.
It’s also something of a bedside backstab. Digital artists use AI tools on the daily; filters, shaders, brushes, layers. The idea that we’ve reached a point that *this* use of AI is too much seems arbitrary. What’s the guiding principle?
I think that you’re right, and the problem is economic. Will the people in creative jobs now still have those jobs in a decade? I don’t know. Maybe. Some. There’s still space for traditional artists. I think that the successful artist going forward is going to have to figure out a way to train AI to make their work quicker, and work in bulk, so there’s probably going to be a contraction on professional digital artists, but who knows? Maybe AI tools democratizes art and there’s more demand for it. AI is awful at narrativizing, so maybe the market will shift to an artist feeding AI stories and directing the outputs.
What I don’t think is going to happen is a return to 2010. I don’t think we’re going to (or even should) shelve the tools out of some kind of an idea about moral purity when it comes to job protection. That’s not how progress has ever worked.
I’m honestly confused by what you think “AI” means. None of that is AI. How are layers AI, for example?
For the record, layers are the weakest example on that list, I’d abandon it if that made the difference. But I think we might just be talking past each other and the problem is definitional – I think you may be differentiating between true machine learning and PCG, which… sure. PCG isn’t true AI. But when we’re talking about labour displacement (which is really what this conversation is about), these are the things that matter. If the argument is that AI is bad because it simplifies or replaces artist effort, then the exact same argument could have been made about all of the above listed and a million other things we take for granted.
Take brushes as the example. If a painter wants to depict a field, they’d either have to painfully depict every blade of grass, or compromise their time against clarity on a spectrum to impressionism. Digital artists can click on the button for their grass brush, and roll the mouse over the field a couple times while the program algorithmically maths in hundreds of brush strokes.
Instead of cross-hatch shading a thousand individual lines, there are brushes that will shade that automatically, adding different vectors and angles the more times you swipe over the same area. There are brushes that procedurally generate very intricate lace, trees, or water ripples. Same base pattern with dynamic shading. Same base pattern with filters (and some filters are explicitly AI). The other good example are things like stretch/shrink tools – the pixels being added during a stretch or the pixels cut during a shrink are mathematically determined.
If you want to make that differentiation – Sure, I accept that. But then the question becomes why is AI bad when those tools weren’t?