On AI art, and art in general
Today's tale is a tale of hubris, of disrespect. But also of enlightenment, and in the end, of love.
Art. Since time immemorial, man has wished to depict beauty, as a way to show sincere appreciation, but also to explore one's own ideal, and put it to paper. Across millennia, the craft made possible by our uniquely advanced hands was perfected. Perspective, ratios, outline to detail, color theory... Those are all painstakingly accumulated strands of knowledge, from those seeking to reach perfection, to evoke ever more powerful beauty.
There was a time during the Renaissance, where technical proficiency had reached such towering heights, that flaws had started to appear. Not out of artistic error, but if you aspire to recreate what you see to perfect accuracy, you are bound to imitate the imperfect as well.
What followed was stylistic explosion. Cubism, impressionism, surrealism, countless movements sought to let technique roam unconstrained by muse or subject. Yet those, too, were bound to crystallize into trend and word. What was once innovative became an artifact, with intrinsic value and neatly filed under clean-cut categories.
I believe there is a certain mistreatment, a lack of coverage as to what occurred afterwards. After being dissatisfied with reality, and disillusioned by the experimental, graphical arts started to coalesce around feelings. Suddenly, a new area unexplored was opening up: the human psyche.
This trend was also made possible by the sharing of information. Suddenly, art would leave private collections, and land onto newspapers, then television, to end up on the Internet. With each step in this exponential growth, more and more potential critics would refine and perfect the burgeoning styles, layering upon themselves.
The line between the Arts started to blur, as animation brought movement to picture, then captured music as well as soon as it was allowed, and all of a sudden otherworldly stories had a medium to exist, with unmatched power.
Storytellers took the opportunity to convey stories with this newly found efficiency. They were met with unmitigated success, across ages and cultures, as evidenced by giants such as Disney. Then, in the US, a cultural shift started to occur.
I could not pinpoint it for sure, but the medium of animation came into contact with the world and wonders of children. Poisoned by prospective profit, and drunk on imagination, animation started to flee the onslaught of cynicism, finding refuge in innocence. Cynicism, evidently, retaliated by elevating cameras and actors as more mature. Not out of respect for reality from which animation was removed, but to wallow in tragedy, to sink imagination into atrophy.
The art of animation as a medium, meant for all genres, like many other things lost, found its way in Japan. Over there, the sheer effort needed to produce animation was addicting. Compounded with a historical desire for escapism, from a motherland bringing construction to collapse every chance it gets, from the pain of interaction in a society moving much too fast. Whatever the reason, modern animation techniques were shunned, and the art of the pen was perfected.
The result is known to all, under the loanword anime. In our culture, it has become associated with the art style accompanying it, the ubiquitous moe, an enshrining of traits of femininity and cuteness. This remaining bastion of artistic expression, embracing all forms of love and lust, came to fill the void created by the retreat of Western animation.
Do you know the tale of King Midas? He who sought gold more than anything, absent consideration for even the practicality of it. Blessed with the power to produce it with a mere touch, he destroyed everything that made it worthwhile.
Well, in the late 2000s, forums started appearing, centered around the sharing of visual arts from new and seasoned artists alike. They were especially successful in Japan, where anime was thriving. With them came a dilemma: how do you make the art accessible on its own merit, without the artist's name tipping the scales? Tags.
Small pieces of metadata, digestible by machine, describing what the image is so that it may reach those whose tastes matched. Setting, traits, colors, characters, pose, expression, everything would be documented into a list of bite-sized terms. The art was surrendered to machine, with the promise of carrying it to loving eyes.
Fast-forward to the 2020s. American franchises have ceased to appear, the creative powerhouse is running on fumes. Increasingly, people of all ages find themselves drawn to anime, especially among the recluse, the nerds. It is no coincidence that the forefront of computer science innovation has been tacitly associated with otaku culture, the two form a sort of symbiosis.
Chief among them are researchers. Computer scientists having broken ground with scaling perceptrons, convolvers, and other methods of deep machine learning. With lots of parallel linear algebra, and weights tuned by picking up patterns from underlying material, you can upscale, regenerate, and even recreate said material from statistics alone.
In order to do that convincingly however, you need tagged images. Data sets where visual was mapped to word, so that machine can learn to fill the gaps. I think you can see what I'm getting at.
It was just a research toy, a utility. Never meant to reach market as-is, if there were any applications none of them were any good. But the void was calling. The masses, hungry for connection yet starving themselves out of fear, needed a substitute.
OpenAI yielded, and out came DALL-E. Unrefined. Undesigned. Type in words, get a picture. The quality was bad, so it seemed inoffensive at first, perhaps even a little fun?
But the void was still unsatisfied. The output needs to be better. People started using it as a substitute for the artistic process it was built upon. Naturally, those betrayed started pushing back, pointing out six-fingered hands and gibberish text.
So the void recoiled. The output needs to be better. More undistinguishable. Surely, the pain that I felt is their fault! The hands need to be perfect. The text needs to be perfect. Everything needs to be absolutely spotless.
The innate ability to tell AI output apart persisted, but became less obvious. Soullessness. Uncanny. Technical arguments faded, leaving only emotion. One thing was abundantly clear: machine had betrayed the trust humans had put in it. Rather, some humans betrayed the trust of others.
A rift, fed by misunderstanding of the self on both parts, started to form. The scientists saw the artists as Luddites, the artists saw the scientists as blasphemous, yet neither could articulate to the other what they felt, for they could not fathom what the other lacked.
Nothing is lost, however, as the existence of AI art gave us an invaluable lesson in real art: what it isn't. Appreciation for the human hand, even in imperfection or outright ineptitude, is now becoming a status symbol. Art is no longer about technical achievement, because AI can do that. We are one step closer to identifying consciousness thanks to it.
One day, hopefully, the sciences and humanities, like Hikoboshi and Orihime, will come to realize they need each other.
Edit: I initially wrote January 2025 as published date by mistake, woops!