The Intercollegiate Student Magazine

A Paintbrush, Not a Painter: Why AI Cannot Create Art

graphic art of an origami robot painting a picture

“Meaning is just another name for expressed intention…”
—Steven Knapp and Walter Benn Michaels, Against Theory

“Freedom’s just another word for nothin’ left to lose…”
—Kris Kristofferson and Janis Joplin, “Me and Bobby McGee

“Art is just another word for the human endeavor to create visual, auditory, or performed artifacts that embody the creator’s imaginative, conceptual ideas, or technical skill…” 

“Artificial Intelligence can now make better art than most humans,” declares the sub-headline of an article by Kevin Kiley that was published by WIRED in 2022 – back before OpenAI blew humanity’s collective mind with ChatGPT, and conversations about artificial intelligence became a part of our everyday lives. Kiley goes on to describe the capabilities of AI-powered image-generating software, such as Midjourney and Dall-E, before reaching a conclusion that ought to shake humanity’s self-understanding to its core:

[T]he most important thing AI image generators teach us is this: Creativity is not some supernatural force. It is something that can be synthesized, amplified, and manipulated. It turns out that we didn’t need to achieve intelligence in order to hatch creativity. Creativity is more elemental than we thought. It is independent of consciousness. We can generate creativity in something as dumb as a deep learning neural net. 

While Kiley carefully adds the caveat that “to connect with a human deeply will always require a Creative human in the loop,” he is careful to distinguish this “Uppercase Creativity” from the “lowercase creativity” that he claims characterizes “most human art, past and present” – and which he insists “is exactly what the AI generators deliver.” This bifurcation of “creativity” enables Kiley to announce that “[s]ynthetic creativity is a commodity now,” as if subjecting meaning-making to the logic of industrial production would be cause for celebration.  

Luckily for humanity, Kiley predicates his claims on a fundamentally perverse – and powerfully incorrect – definition of what art is and is for, explaining near the middle of his essay that people make images with AI tools “for the same reason that humans have always made most art: because the images are pretty and we want to look at them. […] They depict scenes no one has witnessed before or can even imagine, and they are expertly composed.” Kiley thus premises his argument on an indefensible conflation of “art” with “entertainment,” collapsing the distinction between the two concepts into a univocal conception of art as ‘alluring novelty’ – a misunderstanding of art’s fundamental function that can only compound confusion about the real relationship between art and AI. 

Let us start with the last of Kiley’s claims – that AI-generated images “are expertly composed.” In his essay “Music Discomposed,” philosopher Stanley Cavell asks “What is composition, what is it to compose?”  “It seems all right to say, ‘It is to make something, an object of a particular sort,’” Cavell continues. “The question then is, ‘What sort?’ One direction of reply would be, ‘An object of art.’ And what we need to know is just what an object of art is.” Cavell begins by suggesting “a minimal answer: ‘It is an object in which human beings will or can take an interest, one which will or can absorb or involve them’” – but then immediately explains why this definition is too broad, and narrows it: 

But objects of art not merely interest and absorb, they move us; we are not merely involved with them, but concerned with them, and care about them; we treat them in special ways, invest them with a value which normal people otherwise reserve only for other people – and with the same kind of scorn and outrage. […] We approach such objects not merely because they are interesting in themselves, but because they are felt as made by someone – and so we use such categories as intention, personal style, feeling, dishonesty, authority, inventiveness, profundity, meretriciousness, etc., in speaking of them. 

Cavell’s point here is crucial: Art objects are differentiated from other kinds of objects because to experience a work of art as art is to engage, consciously or unconsciously, with the manifest relationship between that work and the subjectivity that worked it into being, between the object and the Other Mind that produced it. This Other Mind is inaccessible in itself (like all Others’ minds), but is unmistakably present as that which the work expresses. By virtue of expressing its creator’s internality, a work of art testifies to that creator’s essential humanity – the now-undeniable (if still-unprovable) fact that the Other experiences selfhood in a manner like unto, but different from, oneself.   

Thus, Cavell, explains:

The category of intention is as inescapable (or escapable with the same consequences) in speaking of objects of art as in speaking of what human beings say and do: without it, we would not understand what they are. They are, in a word, not works of nature but of art (i.e., of act, talent, skill). Only the concept of intention does not function, as elsewhere, as a term of excuse or justification. We follow the progress of a piece the way we follow what someone is saying or doing. Not, however, to see how it will come out, nor to learn something specific, but to see what it says, to see what someone has been able to make out of these materials. A work of art does not express some particular intention (as statements do), nor achieve particular goals (the way technological skill and moral action do), but, one may say, celebrates the fact that men can intend their lives at all (if you like, that they are free to choose), and that their actions are coherent and effective at all in the scene of indifferent nature and determined society. […] Such remarks are what occur to me in speaking of compositions as objects composed. 

If calling an object “composed” is, indeed, a comment on the ways that that object intimates the human intentionality of its composer (and by extension, of human existence as such), then Kiley is not only grievously mistaken in asserting that AI-generated images are “perfectly composed” – he is also directly contradicting his claim, in the first part of the very same sentence, that such images “depict scenes no one has witnessed before or can even imagine.” In Kiley’s account, AI-generated images are expressions, in object form, of subjective experience abstracted away from any experiencing subject – something akin to un-composed, composerless compositions.

The fallacy of asserting an un-composed, composerless composition is not new; in fact, it is the exact same fallacious insistence on “intentionless meaning” that Steven Knapp and Walter Benn Michaels indict in their 1982 essay “Against Theory.” Their essay turns on a thought experiment: Imagine that you are walking on the beach, and you come across a stanza of a Wordsworth poem scratched into the sand. A wave comes and washes the stanza away –  but somehow leaves the next stanza of the poem “written” on the sand, instead. Knapp and Benn Michaels explain that “[A]ll the explanations fall into two categories: You will either be ascribing these marks to some agent capable of intentions (the living sea, the haunting Wordsworth, etc.) or you will count them as non-intentional effects of mechanical processes (erosion, percolation, etc.) But in the second case – where the marks now seem to be accidents – will they still seem to be words? Clearly not.” 

This claim turns on Knapp and Benn Michaels’ belief that whether or not a set of marks qualifies as “a word” is determined not by what those marks look like, but rather how those marks function:

[The marks in the sand] will merely seem to resemble words. You will be amazed, perhaps, that such an astonishing coincidence could occur. Of course, you would have been no less amazed had you decided that the sea or the ghost of Wordsworth was responsible. But it’s essential to recognize that in the two cases your amazement would have two entirely different sources. In one case, you would be amazed by the identity of the author – who would have thought that the sea can write poetry? In the other case, however, in which you accept the hypothesis of natural accident, you’re amazed to discover that what you thought to be poetry turns out not to be poetry at all. It isn’t poetry because it isn’t language; that’s what it means to call it an accident. As long as you thought the marks were poetry, you were assuming their intentional character. You had no idea who the author was, and this may have tricked you into thinking that positing an author was irrelevant to your ability to read the stanza. But in fact you had, without realizing it, already posited an author. It was only with the mysterious arrival of the second stanza that your tacit assumption (e.g. someone writing with a stick) was challenged and you realized that you had made one. Only now, when positing an author seems impossible, do you genuinely imagine the marks as authorless. But to deprive them of an author is to convert them into accidental likenesses of language. They are not, after all, an example of intentionless meaning; as soon as they become intentionless they become meaningless as well.

The authors are thus able to resolve the paradox at the center of their thought experiment, declaring that “[t]he arrival of the second stanza made clear that what had seemed to be an example of intentionless language was either not intentionless or not language. The question was whether the marks counted as language; what determined the answer was a decision as to whether or not they were the product of an intentional agent.” Like Cavell, Knapp and Benn Michaels are adamant that it is the intentionality of an art object – the fact that it is meant to be meaningful – which makes that object art. If the sea makes marks that look like words, those marks still cannot mean anything unless we first decide that they are, to borrow a phrase from another of Cavell’s essays, “expressive of mind”; otherwise, all they can be is coincidence. 

As Knapp and Benn Michaels go on to explain, this thinking directly connects to the questions surrounding artificial intelligence:

If our example has seemed farfetched, it is only because there is seldom occasion in our culture to wonder whether the sea is an intentional agent. But there are cases where the question of intentional agency might be an important and difficult one. Can computers speak? Arguments over this question reproduce exactly the terms of our example. Since computers are machines, the issue of whether they can speak seems to hinge on the possibility of intentionless language. But our example shows that there is no such thing as intentionless language; the only real issue is whether computers are capable of intentions. However this issue may be decided – and our example offers no help in deciding it – the decision will not rest on a theory of meaning but on a judgment as to whether or not computers can be intentional agents.

AI may be able to produce “pretty” images that “we want to look at,” as Kiley says, but that is not sufficient reason to construe these images as products of artistry, rather than mechanism. To do that, as Knapp and Benn Michaels explain, requires one to first decide that the program generating the “art” in question is, in fact, an “intentional agent” – that it has a mind, and subjective experiences worthy of others taking an interest in.  Like Knapp and Benn Michaels, I do not presume to offer an answer to that question, and will only insist that to call something “art” is to assert that one’s experience of that object is predicated on the relationship between that object and the mind that worked it into the form in which it is being encountered. If, however, we conclude that computers are subjects that are in fact capable of artistic intentionality, rather than a mindless mechanical mimesis thereof, then it seems to me that there will no longer be justified in calling such intelligence “artificial.” Until then, however, we humans should instead consider AI to belong to the set of tools with which we make art, for a truly artificial intelligence bears no more responsibility for the creation of the work that it is used to produce than a paintbrush, a typewriter, or a camera.

Ilan Ben-Meir is a Ph.D. Candidate in English at Brown University, studying the relationships between technological progress, artistic practice, and cultural production.

~ Also Read ~

Louisiana State University

The populace is ill-equipped at rigorously filtering through the wildfire of information produced by the digital age.