Art has always had to justify itself with every technological advancement. When the camera came along, painters who would’ve otherwise been hired to paint a family portrait were found unemployed, with a camera operator taking their place. So, painters come up with expressionism. I’ll paint how this scene makes me feel, not just the scene itself. Van Gogh becomes stratospherically famous because we suddenly understood what he was trying to do. Now, with photography, we see it as a complete art form in its own right. It isn’t mimicking paintings, nor mimicking real life. A photograph exists as its own thing. “The camera never lies,” we are told. The painter does, however; they infuse the paint itself with their own version of truth. The line between truth and art blurs the more “handmade” something appears. As I write this on my computer – a wonderful invention that left the typewriter and several transcriptionists out of business (until nostalgia brought them back in the case of the typewriter) – several pop-ups have littered my screen asking me to try some kind of AI assistant. I wonder if the same happened when the camera was invented. I wonder if the early advocates for film photography would stand on street corners with fliers, advertising a portrait in just under an hour, as opposed to a painted portrait which could take up to a week, perhaps longer. Now, generative AI promises the same thing. Dall-E says, produce a picture in a few moments! Drawing and painting are much harder. We even put photography in the much harder camp; waiting for the decisive moment to take that picture requires skill, effort, pre-planning. Going out to take a picture isn’t simple, because we all do it on our phones all the time and few of us are accomplished photographers. It requires tremendous forethought. Sometimes pictures do happen off the cuff, and their energy and vivacity are praised – we might even call the photographer a genius.
Perhaps a Dall-E prompt requires thought. One thing that is often said about AI, is that it is the programmer who may or may not be at fault, not the AI itself. Yet, if you tell 10 photographers to take a portrait of the same person using the same camera, that is, the photographers are given the same prompt, you would receive 20 different portraits. No photographer would approach the portrait in the same way. Some would ask the figure to smile, some would give them a costume, some would take them outside, change the lighting, edit in photoshop. None of this is predictable to a degree. Even with my predictions here, I would still be surprised to see the variation, and it would delight me too. With AI, if you give 10 people the same prompt with the same tools, you would get 10 identical pictures. The art is made by the person, not the machine. This would at first seem to be a failure of creativity on AI’s part, but the point of AI isn’t to be itself inherently creative. I don’t know if in the future, artists will have justified themselves beyond AI. Will we see AI prompters in the same way we see photographers now? Who can say. Maybe we will. I doubt it, because this approach of user error as opposed to machine error is still problematic. What we like about art is that it is made by people. Take Aardman animations for example. One of the first things we note about Wallace & Gromit is the fingerprints that can be seen on the plasticine. If they’d gone to the tremendous lengths that it would’ve taken to brush away the fingerprints of the animators, would we care so much about Wallace & Gromit? Probably, because the stories are fun and charming. Still, one of the first things we mention about Aardman is that we love seeing the fingerprints. What we like about art is that it is made by people.
Empathy is a simulation. It comes from the German (not Greek, as people would expect) Einfühlung. It was coined in the 1870s to describe the experience of art. It literally means “in feeling”. Empathy can be described as a conscious attempt to simulate the emotional experience of another. We know we are not literally taking on their feelings as if they are our own. Instead, we create a simulation of their experience based on our own experiences, what they are telling us, and from all accompanying data. Art is incredible at engendering this empathy, because art is simulative by nature. We immerse ourselves in art, and we love the feeling of exiting the simulation too. When leaving the cinema after watching a spectacular film, I feel this sense of the surreal, this feeling that the film is ebbing behind me like tidal waves with every step forward I take. Often, it’s how I feel I know a film is good; that the simulation has been so all encompassing that leaving it feels disembodying. Leaving it is the key point, here. We know we need to leave. In my few experiences of 3D cinema, I have left feeling queasy. Similarly, immersive cinema that rattles your seat and sprays water, hot and cold air, while a very popular gimmick, is really not to my tastes at all. The last thing I want when watching Dune is to have sand and hot air thrown at me. I don’t want to feel that environment, I just want to experience a simulation of it through the director and camera crew’s eyes. These are important boundaries in art. Fiction, I think it can be argued, exists because of this. If art just needed to be as immersive as possible, we would only tell non-fiction stories. Why bother adding music to a poem, if not to purposefully add another layer of both distance and connection from you and your poetry? Of course, art works at its best when music isn’t just a distancing accompaniment but part of the art itself. We can call this art’s dialectic, then. It is necessarily both distant and close at the same time. It needs to feel real and beyond reality at the same time.
Thusly, I would describe empathy as precisely the thing that illuminates the distance between perceptions. I cannot possibly experience your emotions as you experience them, I can only experience a simulation of them. Therein lies the gap between us all. Does AI attempt, unnecessarily, to bridge this gap? What problems is AI trying to solve? Nowadays, I am given a choice. Sit down and draw something, or use Dall-E. One takes a long time, skill, a financial cost for supplies, critical faculties, and effort. One takes some critical faculties and an internet connection. I’ll pick drawing by hand any day. Maybe it’s just my luddite tendencies. After all, it could be – I’m not ruling that out. But it turns out, I actually enjoy drawing, and that is why I draw. What we like about art is that it is made by people. There can be beautifully made images from Dall-E and Midjourney and other generative AI machines. I see them all over Instagram often enough. The people prompting them are creating incredible images that I wish really existed. I saw one image yesterday of a house that looked like a Jean Cocteau face. It was beautiful and it made me want to live in that house. Still, I would prefer it if I saw a drawing of this house; if someone had taken that drawing, taken bricks, cement, and paint to make that house exist. Something that had taken effort to come up with.
AI is only solving two problems in visual art: cost, and time. These are only problems within art if you are a child who has never been made to wait for anything. The creativity isn’t an issue here because the prompters are creative, not the AI itself. It’s just how long it takes and how much it costs. These aren’t really problems to me because I am an artist, not a capitalist. When I look at a piece of art that has taken me weeks to complete, I am never ashamed or frustrated by this metric. When I price it, I have to convince myself to price it higher, because truthfully, pricing my own work is very difficult, and I only ever feel ashamed of pricing it too low. We are going to have to cope with the fact that art costs time and money, and neither of these things are bad. There are obviously cost barriers in art with material; musical instruments are expensive. I draw with pen and ink on paper because oil paints and canvases cost an arm and a leg. AI does not solve this problem. Will you emotionally connect with Dall-E the same way you will with your first guitar? Maybe. We do have a tendency to anthropomorphise faceless technology (see: your instant emotional attachment to any Roomba you might meet). But my first guitar (a £10 three-quarter nylon string from the middle aisle in Aldi) has flaws in it. Some of the glue seeps out on the edges. There’s a pattern around the sound hole that is missing a drop of ink here and there. Over time, the neck has swollen and warped. This guitar has lived and breathed along with me. This further illustrates my central thesis, though: we like art because it is made by people. Even the things that are indeed made by people but have all its fingerprints removed either become anthropomorphised or remain eery emblems of our own hubris. When I see these @midjourneyaddiction Instagram accounts, I want to know more about the person or people behind them. I want to know their names, what their prompts are, where they live, why they don’t want to make things by hand, why they can’t make things by hand. I want to see the person behind the machine, I want to see where their fingerprints are in the machine. Maybe AI prompters will become artists the way photographers have. We’ll create different AI models like we have with cameras, and we’ll understand AI very differently in a hundred years. Are we going to trust tech billionaires to lead us there?