A RANT ON BEAUTY, OR HOW WE BEAT THE ROBOTS
BY JOHN FLATT
So, like, what exactly is beauty?
Bear with me and ignore the impulse to go look it up in a dictionary and no I'm not talking about the intellectual "oh, this is beautiful because the idea is so goddamn interesting" way either. I mean beauty. You know, shit that's nice to look at.
Honestly, this is a subjective and kind of stupid question. Beauty is the stripper that you fall in love with. Beauty is all those things that they tell you it is in undergrad. Line, form, color balance, hue, luminosity, saturation, value and so on and so on until the sun explodes. A collection of rules. For all the inventions of post modern society and the last 100 years of art we're still fucking talking about it. And for good reason too, it's baked into our psychology.
I should probably start with some disclaimers: no, I'm not against beauty inherently, and no, I'm not decrying anything that doesn't look like it was mauled by a bear. I'm specifically against the sole pursuit of beauty. Beauty should be the sweetener that helps the medicine go down, or the counterpoint to whatever is going on conceptually in something, but sheer beauty, vapid beauty is useless. Especially now.
Q: Well yes, sure, whatever you say, but why say all of this now?
We're in a resurgence of conservative values and conservative artistic material. At best, the people in charge of the United States are a collection of pedophilic crony capitalist lapdogs and at worst, they're a collection of pedophilic nascent fascists, and there's not too many things that right wing movements love more than "ideal forms". A quick glance at the things government agencies are posting these days will reveal callbacks to earlier propaganda, pixelated fantasy southwestern landscapes (to replace the real ones they're ruining with climate change), military decadence and grim faced white men that look torn from some hellish spawn-pit made from Loomis diagrams and comic book heroes. All AI-generated, of course.
The snob in all of us might look at these images and pass them off as inconsequential slop- while that may be true for the time being, it won't be for much longer. AI is getting consistently better at producing the things that centuries of advertisers know work. And if they work in ads they work in products and if they work in products that becomes a lot of people's exposure to aesthetics. Millions of people had Norman Rockwell prints, and digital illustration thrives on Instagram already. With the money of one of the world's biggest economies powering it, AI is probably going to do to culture what photography did to painting (if it hasn't already). We're looking at a revolution in artistic beauty standards that we're all going to lose ifwe don't do something. Honestly, we might still lose it even ifwe do. Beauty is inherently tied to cultural standards and if our culture is slowly becoming addicted to AI nationalist slop ... well that's the whole ballgame isn't it? With those stakes in mind, I propose a third option: maybe beauty just shouldn't factor in anymore? Or to be more clear, maybe we need a new way of doing beauty.
I don't say this to be defeatist, and I'm not even mentioning what this will inevitably do (and has already done) to the illustration sector but if there is a machine that can pump out mass market beauty, what's the point in doing it by hand? And if you recoil at that, maybe the beauty isn't the point but the "by hand" is. When photography shattered the need for painting to be representational it led to some of the most exciting developments in art since the invention of perspective (in the west, anyway). Painters no longer had to depict a thing, but could instead depict concepts, emotions, and the spectrum of human experience that lies outside of the gaze of the camera lens. These pieces were visually daring, and in conservative circles still controversial. What we're experiencing now is the backlash to that expansion of beauty's definition, a popular revolt against the avant-garde in the name of baser instincts and taste. Kitsch has gone corporate. The time comes to take up the blowtorch and angle grinder once again and resist what amounts to a corporation-controlled backsliding of culture in the name of profit. I'm of the opinion that we need to start from the bottom again and reexamine some old dogmas.
Maybe by abandoning all these precious rules concerning form and line and color and composition we free ourselves from the rules that AI has to play by. By choosing to stop playing in the same sport we make our own. Will it be hard? Absolutely. All of art history maintains these systems, but Rome wasn't burnt in a day. There will obviously be detractors, those who cling to the old ways and they shouldn't be shamed for it, but one of the real paradoxes of breaking all the rules is that when you break all the rules you inadvertently create something beautiful. It's important to remember that basically every popular art movement is detested at its conception. Fauvism takes its name from the words of a critic who hated the pictures that movement produced. Fauvism was ugly, until it wasn't. We should only hope to be so brave.
Around 6 months ago I was playing the video game Elden Ring with some friends of mine and we got into an argument about difficulty and fun in video games. Being the artsy elitist prick that I am, I took the stance that for gaming to elevate further as an art form, gaming would have to be unchained from the need to be "fun." To be honest I don't remember most of the rest of that argument as it was done over several beers, but I think it provides a good foil for art as well. If two-dimensional visual art is going to avoid thematic redundancy in the wake of AI, then 2D art (painting, drawing, etc) must once again carve out its own niche.
Q: Alright whatever, but what does all of this mumbo jumbo look like anyway?
Glad you asked. I'm not a theoretician and if I'm being honest there's much better painters, but here's some suggestions:
There is no longer any such thing as a "mistake". That failed gradient, that wonky hand, that piss-poor anatomy in that figure drawing? Those are all things that AI is fundamentally incapable of doing. AI cannot make a mistake because AI has no "Hand". AI is a prediction machine, it only knows what "should" be there. It has no idea what could go there. No matter how you prompt it, AI only has access to the limited mistakes people post on the internet, and those mistakes were left there unintentionally. Even if they weren't, AI has no fucking idea what a mistake actually is. Sure, it hallucinates and adds fingers to hands or breaks a neck or back getting a pose out of a computer generated supermodel but it's doing those things as a flaw. I'm suggesting we do it as a part of the process. If you work digitally or with social media at all, leaving intentional mistakes also has the effect of poisoning the well for whatever bot scraper comes by to steal your shit the moment you post it. Will it look weird and kind of uggo at first? Probably, but use that mistake as a jumping off point to make something new. TLDR: pentimento is more important than ever.
AI has no soul and therefore has no concept of personal color significance. I like red. Big fan, and have been a big fan since I was 3. I can make connections with red that the tag system driving AI would never have any reference for. AI has no idea that the red I love the most is the same shade of red that striates the canyon walls back where I grew up, and even if it has "seen" those canyon walls in a photo, and even if I tell it what exact RGB code to use, there's no way it's going to match the feeling of me mixing it myself. Even if the generated image is technically "closer" to a photo reference, that image lacks the context of me working through the memory of all the things it took to get that shade. These little serendipitous decisions will be behind a meat wall until the bastards figure out how to actually get in our heads, so use the greatest resource at your disposal as often as you can. TLDR: The memories that you have in your head outpace and outperform anything that the AI would ever have access to. Use it.
This is pretty well known at this point, but AI can't actually make anything unique. AI, being a prediction machine, can't make anything that it hasn't seen yet or can't logically predict will happen based on what it has. AI has access to billions of images, sure, but all it does is average them out and then figure out where the next sentence goes in the paragraph (to use an overwrought typing metaphor). How do we counteract this? Make weird decisions. And make them on the page or the canvas or whatever medium you're working with. End lines suddenly, change perspectives mid-page, break rules you didn't even have a plan on breaking. TLDR: "Brute force" your creativity, do weird shit just for the hell of it.
Go minimal or go maximal, AI loses at the margins (for now). This is where theory starts to get baked back in. Walter Benjamin (who wrote The Work of Art in the Age of Mechanical Reproduction in 1935) said that every piece of art has its own "aura". A kind of gravitational pull exerted by its being made and existing in a space. I'm certain that Benjamin would disagree, but I actually do think that AI pieces (at least when printed) have some variety of aura, a rancid and disgusting one, but still an aura. I do feel a thing when I see something that I know is AI. But I'm also certain that no matter how strong that aura is, for the time being we can out-perform it. For all its problems, Abstract Expressionism provides a good example for this. No matter how many permutations an AI does, AI can't get close to the depth of emotion you'll find in a good Joan Mitchell or Mark Rothko. I'm not advocating for a return to color field painting here- far from it in fact- the more we do a thing the better AI gets at doing it. But AI can't match human hands in creating emotion out of negative space- maybe it's a coincidence but most AI art I've seen is actually pathologically afraid of negative space. There's always a pattern or something filling the noise where pure visual silence would be otherwise. On the same token, true over-saturation isn't something most AI models are really good at. Fill the page, leave no border and you're already on terra obscura for the machines.
Of course, all of this could change. Technology is moving at an incredibly rapid pace and is fueled by an economic engine that only looks for profit. This is not meant to be a list of strategies that will always work, but breaking out of any system starts from the bottom and works its way up. Machines adapt. So will we.
