Art is inefficient
Notes on Terminology
I’ll be using “A.i.” instead of “AI” in this post. I got this notation from writer and photojournalist Greg Pak, whose use of the lowercase “i” denotes that what we call artificial intelligence is not intelligence. It also distinguishes the term from the name Al. Since most online writing is in sans serif typefaces, it can be very hard to be a person named Albert or Alvin or Alan right now!
I will also be using “creation” and “create” throughout. I’m not a big fan of the term “creator” as a catchall for “artist,” but I don’t want to limit myself to talking about writing. What I am saying here applies to many art forms and creative modes, including some creative methods that its practitioners may not consider art.
“Get on board the A.i. train or you’ll be left behind.” That message is loud right now, in part because publishers are favoring it. Artists are unquestionably losing ground to the endless, fast, low quality content that large language models (LLMs) generate. If the point of creation is to make a lot of something, human beings cannot keep up with the output of programs that can turn a brief prompt into a full article or image or sound or video in moments.
But the point of creativity is rarely exclusively to make a lot of something. Before I dig in, I think it’s important to acknowledge that not all creation has the same goals. Not everything needs to be a masterpiece, and not everything will be created for the sheer joy of creation. Sometimes the point is to fill a gap, complete an unwanted assignment, send a necessary email to your boss without pissing her off, or meet a deadline. Not every act of creation requires our full buy in and interest.
However, when making art, as opposed to fulfilling an obligation, there are some aspects of creation that remain fairly steady.
Creation feels good. This is often treated as an afterthought, but I’m putting it first because I think it’s actually the central issue at stake. Humans are driven to create, even when it’s against our immediate interests. We like making things! We like making things even though (and partially because) making things has a lot of really frustrating moments. It wouldn’t feel half so good to have come up with the perfect ending for one of my essays if I hadn’t struggled with it.
Creation is a method of thinking. We tend to treat thinking as an internal activity, but a lot of thinking happens through a body in motion or an act of creation. Writers do not sit down to write knowing each word that will follow in order. Even a meticulously planned work is one that will change as it is created and refined.
Creation is for an audience. This is true even of private creations. The audience might be the artist alone, but works are made for an audience, and it is by considering the audience that the artist will determine the scope and aim of the art. Creation is communication, sometimes (often) with the self. It can never say the identical thing to each different audience member. It can’t even say the identical thing to the creator each time.
Creation is intentional. In his essay “Why A.I. Isn’t Going to Make Art,” the writer Ted Chiang defines art as “something that results from making a lot of choices.” These choices are often miniscule and subconscious, but they are choices nonetheless. In writing this blog post up to this point, I’ve repeatedly written sentences and then immediately gone back and changed an article from “a” to “the,” or erased an entire line of thought when I came to see it as outside my goals for this piece. Whether I use “a” or “the” probably isn’t going to be what makes or breaks this essay, but it matters to me because I’m a person.
Creation is conversation. I know I already said it’s communication, but I mean something specific here. When you paint a picture, that picture is based not just on the idea you had, the reference you chose, the landscape in front of you. An image or writing or music or dance is based on what came before it, even if the artist isn’t consciously engaging with that past. We understand things in the context of what came before. I can say that I’m feeling blue and you can know that I mean I’m sad because you know that a lot of English-speaking cultures associate the color with sadness, which is why there’s a whole sad musical genre called the blues and why sad paintings are often blue and why a blue tint to the screen in a movie cues you to feel sad. That’s not a meaning that is inherent to the word or color. It’s developed over time through observation and practice and repetition. Meaning only exists in context.
This is an incomplete list. But even this partial (in more than one sense of the word) list reveals why creation is inherently inefficient. If creation is a mode of thinking, skipping a step, even a boring or frustrating step, is a missed opportunity to think. Outsourcing intention to software that cannot have any means that while the prompt may be yours, what emerges is not. If you did it yourself, you would, with certainty, produce something different than what the software will produce. When you refine it, you’re not refining your own thoughts.
Okay, you say, but I don’t use A.i. for creation. I just use it to brainstorm. But brainstorming is not only a part of creation, it’s an essential part of creation. There are so many engaged and active ways to brainstorm that don’t require you to give your own task over to software that makes you less capable. Just talking to another human being about your idea is going to make your brain spark more than any LLM ever could. I cannot tell you how many times I’ve felt stuck until I sat down and tried to explain how I was stuck to another person.
A.i. is undoubtedly good at making things quickly and in quantity. That’s not why we make art. If we don’t get on board the train, we might be left behind. But it’s worth asking where that train is going and whether we want to go there.