Artificial Intelligence (AI) image models are trained on incomprehensibly vast amounts of training data that is often scraped from the web without any differentiation, resulting in what is essentially mass plagiarism since it is impossible to opt out of the scraping process if you put your work out there on the internet. Cases of AI images with simulated watermarks and prompts that target the style of specific artists reinforce the notion that AI images are inextricably linked to artists that will never be compensated by the companies profiting off AI. This is most apparent in the recent trend of “Studio Ghibli AI” images, which take directly from a singular studio as its source for image generation. OpenAI locks more advanced features behind a subscription model and so rakes in over a billion dollars in revenue. Studio Ghibli is not seeing a dime of that money while being the source for a significant portion of those recent profits. Partaking in the trend, even if you aren’t paying OpenAI via a subscription, is still giving the company free publicity and implicitly endorsing the mass theft that underlies contemporary generative-AI systems. It’s not a good look, and people who know anything about this will look at you sideways.
Now, there is an argument to be had about the emulation of a style. A common retort is that human artists train the same way as machines do. I don’t believe this is the case. The distinction I draw is one of intention. When an artist takes long hours to study the work of someone like Hayao Miyazaki, there are very few people who will study it solely to emulate that style and nothing else. What artists take from the study of greater artists will differ widely. Someone may like the simple shapes, the colors or the representation of nature or the composition, and that is what they take from the study into their own work. Even if you study only to learn how to “draw like Studio Ghibli”, you inadvertently add your own touch by choosing what to focus on, what to draw and how you draw it. AI does not have the kind of discernment that people do. It consumes the work wholecloth in order to make an imitation with modifications grafted onto it. What comes out is the aggregation of averages, statistically modelled to conform to parameters given to it and easily thrown away with a second generation of the prompt. The intensive study of the principles of art, what makes an image beautiful, moving or impactful and infused with the very real experiences of the person making the art can’t be compared to an undifferentiated mass of random noise being smashed together into a simulacrum of something actually made with intention and care.
Hayao Miyazaki is an interesting man, and the clip of him calling an AI animation “an insult to life itself” went viral for good reason, as it is an encapsulation of the way that many people feel about the proliferation of AI-generated images. What I find more interesting, however, is what he says later on in that clip. He states that “we are nearing the end of times. We humans are losing faith in ourselves”. That, to me, cuts to the heart of AI imagery.
On a podcast, Open AI CEO Sam Altman defended this position by arguing for the “democratization of creating content.” I would note, first, the use of content over art. Content is vague and vacuous. When I hear content, I think of the endless scroll of social media, endless short-form videos saying nothing over and over. But whenever I see an AI image, I think to myself, why? What’s the point of this? When did people become so defeated that they no longer believe in themselves, of being capable of partaking in the most human of acts? Other animals don’t have art; it is a uniquely human practice, and now we’re outsourcing it so we can free up our time. For what? What are we doing?
The earliest evidence of humanity is the handprints we left on walls that we made for no purpose other than to show that we were here. We made art: painting what we saw around us with rocks, fur and grass. Yes, we developed skills and traditions and all the gatekeeping that follows, but the fundamental act of creation is a deeply human process.
Everyone I know has written poetry in a moment of sadness, or sung along to a song that they love or done a little doodle in the margins of their notebooks. There is an innate drive to mark our impermanent existence in a world that is constantly changing around us in some meaningful way. Why do away with that?
Art is more democratic than it ever has been. I was in high school with nothing but a stack of printer paper that I took from a school office room and a cheap half-broken number two pencil, and I started drawing my own face in the mirror because I was bored. Six years later, I’m a studio art major, able to make paintings and sculptures, still using the cheapest materials I can find to do so. Art for the past, let’s say — for the sake of argument, I don’t want to get into it — 180-ish years has been about breaking away from elitist and exclusionary practices. Graffiti, performance art, minimalism, the commercialization of art supplies, the inclusion of women and other minorities in the mainstream and so many more developments in art practice have made art more democratic. Note that all of these are driven by people tearing down barriers to make their voices heard and to create art that inspires the next generation of people to keep being human. We do not need machines to make art more democratic; we need a system that values art and what it gives back to us all.
The true democratisation of art lies in fair compensation, an economy that provides enough for artists to create in comfort and in expanding programs to bring art to more people. This faux democracy deprives people of the opportunity of actually creating their own work, instead of asking a textbox to hand them something that looks passable enough. This process kills the artist in people. It’s sickening, really.
