[Posted this in the bad place a few days ago. Let it sit for a few days, now reposting it here with minor edits & additional thoughts. Additional commentary is in square brackets.]
(I have to do the rant.)
So-called "AI art" is a means by which tech capital seeks to plunder value created by workers in creative fields. The rich-who-own found a new way to steal from the poor-who-work. They want to make it legal & unobjectionable. It should be neither.
Point 1, theft
This is mass infringement vs weak legal targets. They know this. Compare so-called "AI art" vs "AI music" that's been extremely careful not to tread upon the copyrights owned by large media corporations. ( ...Whose exploitation of musicians is a separate problem.)
Point 2, it's not AI
These are not thinking machines, they are not creative, they are not conscious, they are not artists. To call it "AI" is a rhetorical deception which obfuscates and romanticizes the issue by invoking images from sci-fi media which are totally irrelevant. These are procedural, statistical algorithms that re-process images - taken without permission from artists aka stolen - into different images. That's it.
[To soften this point: I really just think the field of Artificial Intelligence is a bit romantically over-named, and this gets them into lots of self-inflicted trouble.
Personal connection: My dad worked at the Stanford AI lab and got his PhD there in the 70's, so I've heard a lot about this and about the culture; at least for a certain period. They did some neat stuff; him mostly with computer graphics. They got absolutely nowhere with creating an artificial mind, but they did take some interesting swings at some interesting problems in computing. Of course they were all hippies back then rather than dead-eyed wannabe Silicon Valley entrepreneurs, so their sense of ethics was more developed than the current crop, I think.
Anyway: What "AI researchers" are doing in plain terms is solving computing problems - that may or may not interact with the real world - that require being more clever than simply doing lots of math. (Though it's definitely also that.) Like, can a digital camera detect if a piece of fruit is bad? Can a computer system detect certain unsafe conditions of vehicle operation? Can a website provide the correct FAQ response to naturalistic human-created text input?
For most of these examples: yeah, mostly. But none of these things are consciousness. This software doesn't "know" anything. It's not, in fact, artificial thought in any sense of the word. The field should probably just accept that it is really "cybernetics".]
Point 3, the problem in personal terms
So-called "AI art" won't really hurt me. I have my niche, and it's little-desired by image regurgitation enthusiasts. But I know that one day someone will generate terrain tiles using my assets, and when that day comes I will be pissed.
This would hurt much more if I was just starting my career. New freelance artists rely on cheap/sketchy/unreliable entrepreneurs to get their start. Those people will 100% use art regurgitation instead of paying humans if they can.
It's going to get harder to get started in art.
[Maybe I should temper the line about "cheap/sketchy/unreliable", because a lot of my early clients were... kinda that, but also usually rather nice people just trying to make a thing in, one must admit, the hobbyist sense. I like to think that they'd still hire an artist because they want to make products with some soul in them.]
Point 4, on algorithmic image reprocessing itself
I did a small amount of work studying and building procedural image generation tools in art school in the mid-2000's. It's a fascinating field and intellectually I respect the work being done on, and with, these algorithms.
In school, the moral/legal limits of using others' art was discussed, and boundaries of copyright made clear.
TBH it's very permissive when a real artist repurposes the work of others'. The social cost for such gadflies is low anyway; their art-borrowing is merely annoying to those whose art has been borrowed, it doesn't hyperintensify exploitation of an entire class of worker.
(Aside: writing prompt tags to feed an algorithm is not the same as transformative artistic creativity. Like holding a pencil: yes, it's a skill, but it's trivial vs. drawing. Drawing is a difficult skill and it should be compensated when its output is commercially exploited.)
TLDR
I object to so-called "AI art" facilitating mass theft from art workers. I object to it being called "AI".
It should be illegal to algorithmically regenerate images using artists' work without explicit permission. Especially if that output is to be used commercially.
... Actually, my argument is that it IS illegal. It's just that the law isn't being enforced because artists, as a class of worker, are poor, weak, and divided.
The only way we can fight back is together, in every way we can.
[On copyright: Upon reflection, I might soften this point on copyright: copyright is what it is enforced as, and admittedly it is enforced as something often very stupid and overbroad and mostly on behalf of giant media corporations. So maybe my statement would better be: these so-called "AI art" developers are violating the popular spirit of what copyright is thought to protect, and should protect: the ethics of not stealing people's work.]
[A good example: In the replies, I liked the example of a hunting rifle vs. a machine-gun mounted on a remote-controlled robot. These are essentially the same tool in different forms, but one of these things is legal, the other is not. Society determines whether the way a tool functions is acceptable or not based on its implications - is it dangerous or damaging to society? Is the scale of that possible damage limited? Is it easy to tie a human agent to responsibility for that damage? Is it easy for a human agent acting with criminal intent to cause excessive damage?]
[As with a gun, these questions can be asked about a program that makes copies of peoples' art styles by using their art as a base without permission and without compensation. Do we accept the social damage this would do? I should hope not.]
[The previous line was a great concluding statement, but I do need to comment on the trolls. One was incredibly stupid, and posted a generated image of thick manly christian knight-men that weird reactionaries love so much. Said troll accused anti-"AI art" sentiments to be based on wanting to be paid to draw perverted furry things. ... Which I respect! But that's not art I make, so their reply was a little off-base.]
[Another semi-troll was the techno-optimist that implied that I wanted to censor technological advancement. Deep sigh; look, I like techno-optimism, and it was definitely my primary worldview when I was... um, a child to young adult. But if you're pushing techno-optimism while venture capitalist vampires impose their hyperexploitation programs on us, then you're naively providing a smokescreen for some of the worst people in the world with some of the most power to make our lives miserable. Have some self-awareness! And center the question of: is this technology being used in a way that will increase human suffering? If so, give a thought to why that might be, what the power relations are behind how this technology is being rolled out, and don't be a boot-licker, please.]

