Asking whether AI outputs are art is asking the wrong question

Is this art?

Over the past year, a heated debate has surfaced regarding the use of generative AI, specifically concerning whether AI-generated images should be categorised as art. This discourse has grown quite animated, with one faction staunchly asserting that art can solely be a product of human creativity, while the opposing viewpoint posits that anything (including AI-generated images) can be considered art. Though I am neither an artist nor an art historian, and thus cannot provide expert insight into this matter, I’ve chosen to share a few thoughts, recognising they stem from my personal experiences, not expertise. My intention is not to sway the debate either way, which will likely persist for decades, but rather to offer a glimpse into my perspective as someone who has engaged with these tools for over a year and to suggest that perhaps this is not the right question to be asking at this moment. Feel free to regard the following thoughts as simply personal reflections.

I would first like to start with some background about my own experiences with art. For as long as I can remember, I’ve loved art. As a child, I would spend hours looking at illustrations in books, staring at the details and at the different styles. I didn’t know anything else, but I know that I liked bright colours, and particularly enjoyed old paintings. I also discovered that I had an affinity towards the least appealing artworks; things others would find ugly I would find more interesting. Growing up in Costa Rica, I didn’t have much of an artistic education other than at school, where we learned how to draw and paint, and I have to admit that I was pretty bad at it, but I also found an aptitude in secondary school towards industrial drawing, an affinity that I never exploited. I have kept some of those early drawings, but my aptitudes were in writing, so I pursued the letters instead.

But my love of art never diminished, on the contrary, growing up we travelled quite a lot, and we lived in Mexico and Guatemala, so my artistic tastes evolved to appreciate non-traditional art forms, I particularly fell in love with the art in the Aztec codices, as well as Mayan estelae, and the giant Olmec head sculptures. At this point my parents started buying indigenous art, and I was fascinated by these images, paintings where the perspective was all wrong, people bigger than buildings, dogs larger than lampposts, and all filled with so much colour. I would spend what seemed like hours lost in these paintings (we only had a handful of TV channels back then, and no Internet). I later learned that this art had different names: naïf, primitivism, folk art, outsider art; the main characteristic being that it was performed by people with no formal training. Later I learned to include other similar “outsider” art forms such as kitsch, as well as religious art. The uglier the better (and yes, that includes the dogs playing poker).

In university my artistic connections grew, we had a very comprehensive humanities first year, so I learned how to appreciate art more, but also learned about different artistic movements. I also got to know a few artists socially thanks to my sister who was studying Art, and also got to attend more galleries and art shows. And here my tastes changed a bit, I have to admit that I became more conservative. I hated almost everything I saw in those exhibitions, I found the art lazy and unimaginative. I would complain loudly “but that’s just a black line in a white canvas! I can do that!” “Wait, that’s just someone spraying paint into a canvas!” I even came up with a name for this, “chorreonismo” (roughly translates to “spray-ism”). I would listen to some artist explain their creations in long and complicated essays, and learnt to speak their language for fun. I would of course become outraged by Pollock and Duchamp.

But then something changed. Towards the end of university I really fancied a girl who was in art school, and we went out to a few galleries, and I expressed my disdain for all of the abstract nonsense on display. I still remember some of the conversations quite well, and I think that we never got together because she probably thought that I was a bit of an idiot, but I digress. She asked what art I liked, and I of course mentioned some mainstream stuff, Klimt, Monet, Van Gogh, but I also mentioned the naïf art, particularly Rousseau. She challenged me on several points, but she told me something that stuck with me:

“Everything is art if there’s an artist behind it.”
“And who is an artist?” I asked.
“Everyone is an artist.”

I didn’t agree then, and I still don’t agree now, but it helped me to expand my tastes, so I came to appreciate sharks in tanks, dirty beds, and exploding shacks as art. I still fall under the camp that if everything is art, nothing is art, so I will draw the line at taped bananas, but just about. The thing is that I have found that those pushing the boundaries are always facing gatekeepers, and I dislike the idea of someone getting to decide what is and isn’t art.

Which brings us to artificial generated images and the current debate. When I first saw the first images generated using DeepDream back in 2015, I really liked them, and while I had mixed feelings about the quality of some results, it never occurred to me to really ask the question of whether they were art, I just took them as they were. The thing is that even these early images tickled something in my brain, the same part that loves primitivism, religious art, and kitsch. And as the systems improved, so did the outputs, and the more I liked them. Sure, the space is filled with soulless, derivative, pedestrian stuff. But there are some nice things out there, at least for me.

But is it art? I don’t know, and I don’t care. I think that it’s too early to make a decision, but I find that the debate reminds me of previous ones about new art forms and diverse artistic expressions. So I don’t know if all AI images are art, or none of them are, or just a few made by established artists. Ask me again in 20 years and I may have formed an opinion. However, I do have a stronger idea, and it is that whatever we think about them, there are artists that are using the technology, and those are the ones producing some interesting results, and those would probably fall under the category of more traditional art forms (see here), and it is being taken at least partially seriously by some. In an article in Wired, Raphaël Millière wrote that some of the creations being created are starting to produce curation challenges, a corollary of this is that at least some generative products are deserving of these efforts.

Is there a lot of dross out there? Of course there is, the internet is becoming flooded with an endless stream of anime waifus, but that is not the point. I still don’t think that everything is art, or is even intended to be art, but perhaps something else.

This brings me to my own experiences in generating images. I have to admit that besides enjoying some outputs, I have become interested in generating images myself, and I have folders filled with those experiments, some may have even made it to my papers, teaching materials, and presentations. It has become a nice hobby to experiment with the tools and see what comes out, although my efforts always hit the wall of me having a rather limited imagination. Presented with the entire universe of creativity, I always tend to produce llamas. And cats. Llamas and cats. Llamas and cats, in cyberspace!

But what is going on in my head when I use these products? Is what I do creative? I don’t think so, there’s a degree of technical experimentation that I enjoy, and perhaps brings me back to the enjoyment of technical drawing of my secondary school years. I enjoy playing with prompts, playing with words, it’s like learning magic spells, “abra cadabra, presto!” I guess some could argue that even a level of technical knowledge involves a level of creativity, but I don’t like to think of myself as either a creative nor an artist.

I’ve come up with a different way of thinking about it: exploration. One of the most useful technical terms that describes the process of generative AI is latent space. Imagine you have a huge box of crayons, each one a slightly different shade, representing all the possible colours (or “data points”) you might want to draw with. Now, let’s say you create a simpler, smaller box that only has a few crayons but still allows you to draw a wide range of colours by blending them together in different ways. Latent space is like that smaller, simpler box of crayons in the world of data and artificial intelligence. It’s a reduced, simplified space where complex data, such as cats, or llamas, can be represented in a more manageable way. In this space, variations among data points (like different facial features) can be smoothly transitioned from one to another, just like blending different crayon colours together. This allows computers to understand, generate, and modify complex data, such as images, sound, or text, in a more efficient and intuitive manner.

So what I think I’m doing with prompts is exploring latent space, and producing an image is a discovery of a combination of words that produces a result. So this is not necessarily art, or creative, but a discovery. I’m not sure if anyone else will like this analogy, but it works for me. Take the image below. I really like it, but it came about purely by accident, I was playing with Midjourney and typed: “the unbearable lightness of being, with cats”.

So it’s not creativity, it’s not art, but it’s something else. A discovery perhaps. I discovered a corner of latent space where this image came into being. I like it, and that’s all that matters in the end, your mileage may vary.

Concluding

While this post is not about the law at all, the question of what is art has actually come up in legal opinions. One of my favourite copyright cases is Hensher v Restawile, the details of the case itself are quite pedestrian, it involves a sofa collection, and whether the furniture met the legal definition of artistic craftsmanship. The House of Lords had to analyse in detail the question of “what is artistic”, and came up with a delightful discussion on that topic. I won’t go into any detail, but they went through variations of the above debate, from “everything made by an artist with the intention of being artistic” is art, to gatekeeping ideas of something that is pleasing to the eye and recognised as art by the majority of the public. The view of intentionality in art has prevailed later, such as in the fabulous case of Lucasfilm v Ainsworth. I know what you’re thinking, “Andres, don’t bring law to an art fight”, but I think that everywhere we look we continue to struggle with the question of what is art. So if judges and juries can’t agree, and art history experts can’t either, perhaps we shouldn’t even try it.

Now, what do I get if I type “a painting of a llama made in naïf art style“? There’s only one way to find out…