As a writer, I haven’t touched ChatGPT and, truthfully, I don’t plan on using it (or any inevitable variations to come). I used to mess around with some of the AI art generators (Artbreeder) a few years ago when the outcome was more dreamlike and generally incoherent. Those models were mainly trained on photos of animals, landscapes, and paintings—which made what they spit out a collage of faces, fur, and oddly psychedelic images. What drew me in initially was the way the images fell within the Uncanny Valley.
In those images there was something particularly frightening and, as the Youtuber Solar Sands talks about in his 2020 video essay, ‘Can You Name One Object In This Photo?’, “[AI art] is in this weird middle ground between complete abstraction and representational object.” This was over three years ago, and AI art has improved so much in that time, some people are impressed by the way it creates new works.
Though, as I’ve noticed, it’s not the art world that is impressed, but it’s the people who see art as purely a means to obtain more capital (See Hayao Miyazaki’s reaction in the sources below). One example is the graphic novel, 'Zarya of the Dawn', in which Kris Kashtanova fed prompts through AI. Kashtanova works at an AI company, with their website, Instagram, and Twitter solely focused on AI art as well. There are debates on the graphic novel’s copyright, but what makes me uncomfortable is the way it’s seen as the new wave of art. These AI models unethically source images and text without consent from the original artists, which is another problem altogether. Many other examples exist where the intention of creating a product isn’t aligned with creating art.
Even then, I’d like to focus on a scenario, possibly not far from now, where the databases are ethically sourced, the person feeding in prompts has copyright, and the AI art is indistinguishable from human created art. Also, for this instance, let’s assume whatever novel/short story/poem/painting/photo the AI creates has a cognizant plot, characters, description, and is logically correct. What happens then? What kind of readers would be interested in this work? Is it worth reading a piece that a person had no say in its creation? Well, we know the AI would be able to create works at a rate much faster and cheaper than humans. Thus, there will be a glut of work being produced, pushing out artists simply due to the demands of capitalism. But it’s not just the loss of jobs, but rather the detachment from human experience is what I think would make the art indigestible and hollow.
It reminds me of the legend and website, the Library of Babel, where there is a library comprising all the combinations of words there could ever be. Somewhere in the library there are all the works of Shakespeare, Stephen King, and yet undiscovered literary gems. Though, because of its vastness, there is virtually no way to find what you’re looking for. This is what I think AI art would become, in that there will be shelves packed full of words that have no other meaning than themselves. And particularly, AI cannot capture what isn’t said—what information is withheld intentionally.
I engage with literature, not because I want to know what an AI thinks about the human experience, but because I want to know what another living person has experienced, interpreted to be meaningful, and gave to me. An artist doesn’t make art because they want to make money (though, that’s a major plus), the same cannot be said for the people who tout and consume AI art.