Art Pollution: What Is AI Doing To Our Sense of Beauty?
During college, I made regular visits to the Barnes Foundation in Philadelphia, one of the most impressive modern art collections in the world. I was a budding writer, not a painter, but I was taking my cue from Hemingway, who claimed to have learned much of what he knew about writing from the paintings of Cézanne.
Speaking of Cézannes, the Barnes boasts 69 of them. It also has 181 Renoirs, 59 Matisses, 46 Picassos, 16 Modiglianis, and a slew of other masterworks. The collection is magnificent, but it’s also a lot to take in all at once. At times I have even felt that it was too much and have warned friends planning their visit to prepare for sensory overload.
Now if I can feel that way about a 23-room brick-and-mortar building, what am I to make of the Internet in a time when AI is churning out many millions of images a day?
Granted, the comparison is silly. A webpage might be an apt analogue to an art museum, but the Internet is more like a city, a planet, a universe. Still, is there a point at which we might deem even a whole universe polluted by art?
Many commentators have observed that new tech like deep fakes and large language models such as ChatGPT threaten to imperil our already-tenuous hold on objective truth, but I haven’t heard much talk about what AI image generators like Dall-E, Stable Diffusion, and Midjourney might do to our sense of beauty? How will our perceptions of authentic, human art change once the public square is awash in technically proficient frankenart?
As far back as 1973, John Berger, in Ways of Seeing, wrote that, “For the first time ever, images of art have become ephemeral, ubiquitous, insubstantial, available, valueless, free.” I can only imagine what he’d make of the artificial future we’re steering toward.
Granted, art has always been in some sense artificial — it’s no accident that those words share a common root — but until now, regardless of whatever tools we may have manipulated with our hands, the human mind remained in the driver’s seat. That’s no longer as true. Humans may still be directing the machines to some degree, but not even the engineers who designed those machines can say what goes on in the “black box” of their neural networks. I have to wonder: Will there come a time when image generators are doing their deep learning on a corpus of work that consists largely of their own creations? And where will that leave us?
Art is already a second-order reality, a representation of the world, the “mirror up to nature,” as Shakespeare put it. Art reproduced on the internet puts nature at a further remove — let’s call that third-order reality. Art produced by artificial intelligence by pastiching the collective output of centuries of human artists takes us out to a fourth level. If some of that art it trawls for influence is its own creation, then we’re at least five levels out. Whatever may or may not be true about the ultimate nature of reality, many of us certainly will live in a simulation — a funhouse-mirror version of what Baudrillard called the “hyperreal” and Morpheus in The Matrix called the “desert of the real.”
I’m old enough — forty-five — to have grown up in a time when scarcity still felt like the rule in art. On those occasions when I was allowed to get a new comic or magazine from the aisles of magazine racks at the grocery store — the closest thing we had to the Internet back then — I’d pore over every image until they were burned into my brain, then I’d rip out half of them and tape them to my bedroom wall. I loved those images, and can still instantaneously call many of them to mind.
Now, I live in an age of overabundance, with many-splendored images competing for my attention all day long, and I rarely pay them much attention. Maybe this is because, given their easy transmissibility, they’ve lost their “aura of authenticity,” as Walter Benjamin put it in “The Work of Art in an Age of Mechanical Reproduction.”
Or maybe it’s just that beauty too is subject to the law of supply and demand. I take it for granted that I will see a million more interesting images tomorrow, so I don’t need to linger long on the one I’m beholding now. Or now. Or now.
For most of our evolutionary history, fear of scarcity impelled us to go hunt or gather that next meal — our survival literally depended on it. Then, around 12,000 years ago, we discovered agriculture, stopped wandering, and started accumulating. Anarcho-primitivists see that moment as humanity’s original sin, the beginning of so many of the ills that plague us today. I think they’re probably onto something, though I’m enamored enough of modern conveniences like the grocery store that I’m not sure I’d undo that discovery if I could.
On the other hand, the way we’ve transferred our mania for abundance from the realm of food to other cultural domains such as art strikes me as less obviously a good thing. We’re hard-wired to find certain experiences rewarding. Anyone who has ever had the pleasure of discovering a rare, sought-after book in a second-hand bookstore understands the hunter-gatherer thrill that can accompany scarcity. Mary Poppins said, “Enough is as good as a feast”; in the case of art, I wonder if enough isn’t actually better than a feast.
For the anthropologist Claude Levi-Strauss, the purposes of art included: 1) expression of the imagination, 2) ritualistic and symbolic functions, 3) communication, 4) entertainment, 5) political change, 6) social causes, 7) psychological and healing purposes, and 8) propaganda or commercialism. In theory, there’s no reason AI-generated images can’t serve most of those functions, but the cynic in me (or is it just the realist?) assumes that, once our current gee-whiz fever has passed, propaganda and commercialism will rule the day. The Internet, already so awash in dreck and misinformation, will become a nearly unnavigable hellscape of interesting images devoid of history, context, meaning.
I don’t want to downplay how awesome these new art generators are. I’ve had some fun experimenting with them myself, and I’m all for democratization in general. I’ve also witnessed more than a few intrepid artists collaborating with these AIs to create works of astonishing originality. Still, that I can’t immediately think of a specific one to hyperlink to feels like it might mean something. I suspect that, as the sum of cyborg art on the internet approaches its asymptote, my level of engagement with it may plummet toward zero.
So how should we respond to art pollution?
One answer might be to surrender and simply recognize that art has always been changing in lockstep with technology, and que sera sera.
Another answer might be to remind ourselves of the value of old-fashioned, second-order, human art, and to cultivate our engagement with it, consciously elongating our perception to counter the inundation and flickering of the mediascape. Go to an art museum and study a painting for way longer than feels natural. Spend an hour with a pencil and a blank sheet of paper and see what happens. Dust off an old film camera and take some photos the old-fashioned way (I recognize the irony — to painters in the nineteenth century, the camera was the bogeyman). And let’s not leave out the other arts. Turn off your phone for a day and commit to reading a novel cover to cover. Close your eyes and listen to a whole record. Memorize a poem. And while you’re at it, consider meaningfully supporting artists by buying their wares and through channels like Patreon — unlike AI, creative humans need to eat and pay their rent.
Yet another answer might have to do with recognizing that even this stuff our non-artificial intelligences are engaged in right now — this abstracting, schematizing, languaging — also distances us from first-order reality. All this time, while we wreak our luxurious havoc on the environment (art pollution is literal too), nature is out there doing its ordinary, miraculous thing. Go be in it. Better yet, go be it.
If you enjoyed this piece, please consider clapping, commenting, sharing, or buying the author a coffee.
