.png)
1 month ago
5
The existing discourse around the use of AI in art needs no introduction. Developing from isolated Reddit threads to Reddit communities, from YouTube videos to dedicated YouTube channels, the internet has made it clear that one of the Great Questions of the post-pandemic age is this: “Is AI art Real Art”?
Of course, to anyone familiar with said discourse, the question is fundamentally subjective. It relies on one’s definition of Art and Artist, and at times it seems completely pointless to discuss anything except which of the two prevailing frameworks is the most valid. However, I believe both these frameworks miss out on a crucial insight about the nature of Art and Artists, and I am certain that understanding of this insight is invaluable to achieving any sort of consensus on the topic. Before that, though, I should begin by (briefly) outlining the two existing world-views that characterize the sides of our debate.
Press enter or click to view image in full size
The Sides of the Debate
The first is Art as Process. This can be summarized as the idea that what is valuable about art is the amount of effort put into it, that what is needed from a piece of art is the knowledge that someone spent real hours of their limited life to build up the skills necessary to create the thing that we now see before us. The fact that only so many people in the world can paint in some specific style makes it unique, and it being unique makes it precious. Proponents of this view feel cheated when they are made aware that something they enjoyed was made by AI; they believed that this image of their favorite character, or depicting a struggle they relate to, was made by a human who cared enough about this thing to spend hours or years to create it. When they are told this vision of a kindred spirit is a lie, they are less than pleased.
The other side of the argument favors the concept of Art as Product. This approach prefers valuing something for what it is, rather than focusing on the path it took to get there. If I see something that makes me feel joy, the joy is just as real no matter the source of the thing that made me feel it. If anything, the fact that no one had to work themselves to the bone for me to enjoy my silly images is a boon, surely; lowering the barrier to entry for creativity can only lead to more creativity in the world, can only lead to greater quantities of the things we love. This is the promise of AI, what makes it magical; what if you could have everything, and pay next to nothing for it?
Both views have their merits. The emotions we feel when we experience something do in fact have value regardless of their source. Likewise, some of those feelings are often inextricably linked to our perception of the artist, and this should not be discounted. There are also more nuances to both sides, and I have hardly done them justice; nor will I attempt to do so here. The other parts of the debate are often more disconnected from each other, with talk of AI slop and artist elitism muddying the waters seemingly without purpose. But there is a concise way to articulate much of this confusion. The conclusion that the discourse must eventually lead to, I believe, is the following: Art is not just a Process or a Product, but a Language.
Art and Communication
The concept of Art as a method of communication is far from novel, but it has unexpected implications when applied to our modern context. To understand this, we must consider the nature of communication in general, followed by the nature of artistic communication, before we can finally understand how AI fits into the picture.
Language and conversation is a topic worthy of several fields of study. For our purposes, we will begin with the theory of deconstruction, which we get from the French philosopher Jacques Derrida (among others). This theory postulates that all meaning is deferred, that things always mean something only in relation to something else, and that this means that “no meaning is complete and total”. The word “chair” cannot possibly encompass everything that we might consider a chair in your mind or mine, nor does it detail everything one could possibly know about the nature of chairs, their history, their atomic makeup, etc. It is simply a useful abstraction that lets us interact with the world in approximations without wasting time on details.
This applies on an individual level as well as the cultural. If I tell you I love cats, your understanding of the word will never be identical to mine, even if we had spent every moment together since birth. What kind of cats, what size, what color? We must use more words to convey that further meaning. Of course, this is true all the more for the fickle subjective things that inhabit the most personal corners of our minds; our beliefs, our emotions, the stories we tell ourselves about the world. These things cannot be conveyed with any number of words, not truly; but perhaps we can approach some semblance of that communication by making Art about it.
Art is not, of course, a purely visual medium. It can be an image, a song, a collection of words, or any combination of the above. What makes it special, however, is not the combinations of those things that we see in our social media feeds, but the space that exists between them. If I were to tell you that I was anxious, you would not really understand me. But if I were to tell you that I have not been able to catch more than a few hours of sleep for many nights, that my nails are jagged with bite marks, that my stomach is filled with dread if I allow my mind to wander to the future even for a moment — then you might begin to understand what I have to say.
This occurs because by not stating outright the thing that commonly denotes what I am talking about, I am preventing your mind from using the abstract shorthand that you are already familiar with that would describe my situation without really conveying it. Instead I am forcing you to work for that understanding, and if I have set the stage correctly, the understanding you attain by trying to fill the gaps I have left with my words will match my experience far more truthfully than any direct description could. This is the magical process that we most often describe as Art; the process of conveying meaning in something other than the most direct way, thereby communicating something more accurate to the Artist’s intent.
I should note here that while I have been using emotional examples so far, I do not believe that Art requires some kind of Message. The same principles apply to all things we might consider art; a painting of a beautiful lake, or a photograph of a city street, all function through the same method of drawing your attention to something without fully describing it. I cannot tell you about the lake; I must show you a distortion of it, and through understanding the imperfect reproduction you come to view the lake as I did. I cannot simply tell you that the street is busy with people from all walks of life; I must show you the whole picture, and allow you to discover that truth for yourself as you observe all the details of my photograph and piece them together.
This is the “Soul” that anti-AI proponents often talk about. It is, I think, an appropriately vague word for a fairly vague thing. The aspect of it that is most important to us, however, is what I said a couple paragraphs ago — “If I have set the stage correctly”. No matter the medium, to create something meaningful requires skill, and a high degree of control over every aspect of the end result.
The control is crucial here. It is difficult already to convey something for which there are no simple words; it becomes outright impossible if our control over our work is limited. In fact, we do not need to look to AI to see how this impacts Art; we need only to consider things created by groups of people organized in different ways.
As an example, consider the humble movie. While we often consider certain movies Art, we also often talk about a single movie as several works of Art. We talk about a specific actor’s performance, or compliment the writing of the movie’s dialogue. When we talk about the movie as a whole, we most often attribute it to the director — the person with the most hands-on control over the creation of the Art. Whichever part we consider, the Artist is the person who has the control and creative freedom to decide how that part is presented to the viewer. Likewise, we often talk about “soulless” movies made by large corporations, movies where the artists’ control over the final product was in some way limited significantly by external factors. This, we often agree, kills the “soul” of the movie.
The Implications for AI Art
Having considered the above, the relevance of AI is obvious. The very nature of “generating” content by algorithmically analyzing and synthesizing millions of existing works takes control away from the artist. If I ask Midjourney for a picture of a cat, I do not control what existing images of cats were used to create my masterpiece. No matter how many words I use, no matter my prompt structure, there will be crucial aspects of the end result that I do not have the option of deciding on. The very promise of AI is that less work is required to turn my imagination into reality, which necessarily implies that less control is given to me in terms of how that reality is conveyed.
Despite this, it should be noted that a lot of the output of an AI model is very much within the prompter’s hands. Many tools and best practices are being developed this very moment to improve that control, and one could argue that they represent steps to make AI art more into a “real” art form. However, these steps must be taken with the above in mind; that a primary feature of Art is the ability to convey things without saying them directly, that work is required by both the Artist and the observer for that feature to really shine.
Personally I think that to truly use these mathematical wonders to create something meaningful we must move away from the “consume all the art in the world” methods of training them and instead focus on more modular ways of encoding and reproducing meaningful things. Perhaps a small artist community could create a set of algorithms that encode their unique way of doing things, and combine them in ever-more-complex patterns that create soulful Art unlike anything we’ve seen before. Perhaps an artist-programmer could write code to repurpose their childhood photos into a cohesive vision of their upbringing, and convey it better than any painting or story could. Throughout, the focus must be on giving the artist more control over every part of the process. The possibilities are endless, but I do not think we can push the boundaries of artistic endeavor by asking ChatGPT to generate us scantily clad women with cat ears. Do you?
