Thing I was just pondering: once artists start seeing GenAI as a thing that can make money for them rather than take money away from them -- and can square its so-called "theft" with their ethics, which I do not think is at all difficult to do -- I suspect they'll be overwhelmingly for it almost overnight. (Note: all of the below ignores the environment argument entirely; I'm more focused on the morality-specific objections here.)
3d printing will likely be involved in that shift, as it can generate consistent product (with infinite possibility for tweaks and variations) based on an artist's actual work, and with good enough printers they effectively print product that requires little more finishing than connecting different parts. Merch tables at conventions and the like are already blowing up with 3d-printed figurines and trinkets (a bunch of which are AI-generated, btw); it'd be better if, instead of folks selling the same free Thingiverse models at every table, artists could sell unique pieces based on their art. Hell, they could even do collabs with sellers who have high-end 3d printing setups.
And yes, folks will argue that an artist could partner with and pay an actual 3d modeler to sculpt the models from scratch (or learn modeling in addition to their other digital art skills and spend countless hours doing that instead of the other art tasks they find more engaging), but here's the kicker -- just about any modeler who'd charge a sufficiently low price for the project to be worthwhile *is going to use AI to generate that base sculpt*. (That's just the workflow. Even in the toy industry, AI or some nigh equivalent procedure is so baked into the production process that they don't even think of it as that: high-end modern 3d scanners, for instance, which they use to generate so many actor and wrestler face sculpts, *use AI*. And I still think that while digital art tools may not technically use AI in certain instances, they do sufficient "work" for the artist that I'm disinclined to call selecting a specific filter from a menu and tweaking settings before applying them all that different from asking Whisk to alter an image to specs that match the output of the filter.) So unless the artists ultimately allow for the use of AI, these 3d representations of their artwork aren't going to exist. And maybe they won't exist, because maybe some folks' stances really do derive from well-considered principles rather than protective self-interest. But I'd be more inclined to bet that at least some faction of artists will reverse course and start championing AI in at least limited use cases (e.g. to create merch based on one's own original work, where one isn't so much depriving other artists of work as augmenting one's workflow and diversifying one's table offerings) because they will see it as a benefit rather than a competitor.
So that's the above. But also, tangentially (this was intended to be a quick parenthetical and spun out into the below):
Does the theft argument hold if one uses GenAI to produce works in one's own style? I mean in theory it should, given that the training process that produced an image included analysis of lots of art from different non-consenting artists, and yet I have a hard time saying that an artist who uses AI to produce replicas of their own work -- or, even less, to tweak their own work -- is stealing. Granted, I'm not sympathetic to the stealing argument in general, but this still seems harder?
What gives the theft argument any meaningful heft to my mind -- which is backed up by the specific examples that people cite -- is the suggestion that in any one instance it is potentially harming the brand and/or livelihood of specific artists. The Ghibli-style memes are bad because they bastardize the signature look of those classic anime offerings, which might ultimately make people less impressed by and appreciative of Miyazaki's art. I read an article about an artist who claimed that her commissions were going down because of AI, with the implication that people were copying her style or producing similar images and therefore no longer needed to contract her for unique pieces. That's bad because that artist ostensibly lost money because of AI. (The article was pretty light on details -- I wondered what made her so certain AI was the reason for the downtick in commissions, what with rising prices and all happening these days. A lot of the same people buying commissions from convention artists have other nerdy hobbies, and prices on those have gone WAY up thanks to tariffs. If they're really more into collectibles, it'd make sense for those nerds to spend less on 2D art commissions. But if the artists also sold figurines of their work...) AI "artists" selling AI-generated prints at conventions or exhibiting AI-generated images in an art gallery or posting AI-generated images in artist spaces are diverting money and attention away from real art toward their low-effort slop, potentially harming the real artists who put real effort into their real work. These theft/harm arguments make sense to me.
But somehow it loses something from me when you're stealing lil teeny tiny pieces from everyone and everything? That's not theft; that's just how knowledge and inspiration work: we are constantly "stealing" from everything we've ever encountered, and nobody expects us to even attempt to compensate or even seek consent from that infinity of sources. AI is a little different in that there's actually someone profiting (in theory) from our interaction with this repository, but I'm still less inclined to call this latter characterization theft in any meaningful sense. We need to be able to point to actual people and say that they were harmed, and it feels like a more compelling argument to me if those harms occur in specific use cases that we can rightly condemn. (Or not, depending. I'm not at all bothered by the Miyazaki memes; that to my mind falls under fair use/parody and isn't taking a cent from the Studio Ghibli coffers.)
But "all" artists (and so many other contributors; I think the argument weakens further when one considers that much and arguably the bulk of the training input is not and was never regarded as "art" proper) aren't being directly harmed in the collective argument; that their data ended up part of these projects is the price of the potential to be seen online. I have 20+ years' worth of artwork and original writing online; that content was likely scraped to train GenAI models. How does that directly harm me? Why should that outrage me? And if that knowledge shouldn't -- or *doesn't* -- leave me incensed, and you fail to convince me that mine is not in fact the correct response, why should I be outraged on your behalf?
To say that GenAI harms people by existing -- because certain use cases are unambiguously harmful to individuals -- and therefore even seemingly innocuous use cases are bad is effectively an Original Sin argument. And... okay, fine, but I don't find Calvinism compelling either.
