strongly considering becoming one of those people who insists on never referring to a particular thing by its common name for political reasons because framing big probability models as "artificial intelligence" and pretending they're anything like HAL 9000 or Data is doing so so much damage to laypeople's understanding of a technology that primarily exists to make their lives worse
I genuinely think this is important to do for art criticism, cause like, all the precedents for "AI Art" are actually things like the stochastic methods of the surrealists and dadaists. there IS precedent within art history, but it's stuff like Jean Arp dropping scrap paper onto a canvas.
maybe it's a niche concern, that so little of the discussion around this stuff seems informed by anything from the last actual century of art production and criticism, but it feels to me like it really distorts the conversation when instead of using a term like "algorithm art" we're using "artificial intelligence". it's just incorrect, in a way that feels designed to insulate practitioners and tech pushers from questions like "what is your actual art practice? what are you contributing to the process? why is your work so much less interesting than your average Bob Rauschenberg assemblage or Max Ernst rubbing?"
this is huge. in general, i think vanishingly little "AI Art" is interesting or good as art in particular, but that its taking assets from traditional medium artists has little to do with WHY it's bad. In general, a lot of the anti-AI art takes remind me a LOT of both a) people criticizing sampling in music, from the perspective of like "its stealing to base your song off of an amen break" and b) the freakout during the 90s about how photoshop filters would replace artists.
both of these are understandable objections but misguided from an art criticism perspective. Its true that Gregory S. "GC" Coleman, drummer for The Winstons and thus the person who inadvertently provided the source for everything to Straight Outta Compton by NWA to...at least 25% of jungle and drum and bass music never received any royalties for that sample and died homeless in 2006. That sucks! He should have not died that way. But i don't think people these days would necessarily think NWA or Goldie or, hell, the theme to the powerpuff girls have much to do with the drum solo from a semi-obscure B side that was also apparently written in less than 20 minutes. In all of these cases, at least in my opinion, the work is transformative: referencing art from others, but making it different and holding a certain kind of deference to its source material. All of the above music sampling the amen break is in conversation with the winstons, in a sense. Hip Hop music wouldn't have been possible without those drum breaks, and the discovery of the utility of those breaks was discovered by none other than a high school aged DJ Kool Herc, who premiered the use of this technique using two copies of the same record to keep the break going seemingly forever so that live MCs could rap over it, in a way both indebted to the R&B and Funk songs Herc was sampling but also the soundsystems from Herc's native Jamaica. This technique was amazing to others in the Bronx, many of whom also had connections to the carribbean. Similarly the D&B and Jungle music that so features drum breaks is itself indebted to both hip hop but also another outgrowth from those Jamaican soundsystems: dub, the first real genre based around electronic/mixer based remixing of existing songs.
I don't think a lot of people who take huge issues with the use of living artists' source materials for ai art would have AS MUCH issue with the process described above, even if its just as much about mining art history for samples and not paying its sources, along with many other beloved examples in music (Paul's boutique by the beastie boys, since i left you by the avalanches, chill out by the klf, endrtroducing... by DJ Shadow, Donuts by J Dilla, and, hell, even all the mouth albums from Neil Ciceriega). Why? Because all of this is operating within the principles of what makes modern (ie post 1850) art good when its doing this kind of thing: maintaining reference to the original while still being different from it and also additionally deferrring to it, not hiding from its forebearer.
and this is where AI art largely is different. There is no attempt to be deferential or even referential, no attempt to place ai art within the context of post-1850 art history. I think its VERY EASY to imagine very good art made with ai, but all of this art would not attempt to hide where it came from, what it was taking, and how it changed it. even if it didn't pay the artists, it would attempt to be in conversation with the art that influenced it. Even though that's not something an ai can control, guess what! Herc's Technics turntable wasn't the one who came up with looping drum breaks like amen, brother or funky drummer, DJ Kool Herc did. TNT by Tortoise was not a product of ProTools, it was assembled in ProTools by the members of the band tortoise, who used that tool to achieve creative aims. I have yet to see much of this kind of creative arrangement by those using prompts to generate ai art with the intent of it BEING art. There is no attempt to view the AI as a protools or AKAI MPC-like tool that facilitates the sort of point of view that makes artistic creation, well, artistic. its just used as a Free Art Button.
This brings me back to photoshop filters, and the freak out that they'd replace artists. This, uh, did not happen. There are several reasons for this, but the big one i can think of is that the sort of person who thinks that photoshop filters can replace artists is generally not themselves good at using (or knowing how to hire) people who can use photoshop filters in a way that looks good consistently. They might learn a couple quick tricks (hey, this makes it look like what a random 40 year old's idea of andy warhol is!) but after a couple uses of this it becomes old hat, and everything just ends up looking samey. and, again, the person with the right point of view totally can make something completely transformative in photoshop, combining things they didn't make in novel ways, using filters super creatively, but this sort of person is already approaching the tool the way an artist approaches a tool: as a conduit for realizing existing creative ideas, not as a way to do creativity for you.
and, finally, the money issue. yes, it sucks that ai art can feel like plaigarism that costs people money, but this is an issue that's less the fault of an algorithim and more one of art not being treated as a public good. The (admittedly quite ambitious) best solution to this in my opinion is to, as a society, just pay artists, like we did back with the WPA. the societal benefit of an artist's viewpoint becomes limited when said artist's creative work is only sustainable when its a financialized product, when you become concerned with selling it, when the experimentation of a Bronx teenager building off of a musical tradition bigger than him ends up with a drummer who inspired it dying penniless. us not wanting to pay artists is a political and societal problem, not one of individuals Not Paying For Art Because They're Bad.
But i guess where this leaves me is with the opinion that in theory, AI Art could be very good, but if and only if it is used the way sampling in music is when it is used well: as another tool in the artist's toolbox, that serves to facilitate translating their existing point of view into art rather than as a way to create art wholesale. This viewpoint necessarily involves conversation with the art that's mined, and involves transcending mere reference but also deferring to the original and differing from it slightly.
I have still only seen one person (ai_curio, with their cybersaints project) who appears to be competently using image synthesizers as an artist's tool to achieve a desired outcome; they're an artist who's too disabled to draw what they want, who is going for something very specific and intentional, in creating a body of work with a consistent theme and visual characteristics.
The vast majority of other output I've seen from people working with these tools is not attempting to use the unique characteristics of the tool to their advantage; they're trying to obscure or bypass those characteristics. Much of what is being touted as The Future of Art appears to have little or not artistic intent at all, and no understanding of or attempt to work to the strengths of the tools and medium. The more the latest generations of AI art tools try to be photorealistic, the less character they have.
Their most vocal advocates mean it's "the future of commercial art;" when they say it'll replace artists. artists are more than technical skill. the vast majority of AI art online today, including the stuff being pushed hardest by web3 type people, is bad art. I'd say it's the equivalent of a Thomas Kinkade painting or Peter Lik photograph, but those people know the strengths of their mediums.
woof, this nails a lot of the stuff i've been thinking about but had no idea how to put into words.
