AI Readymade: The End of Art that asks What is Art?
* * *
Innovation in the field of artificial intelligence has reached a stage where two important problems of analytic philosophy can finally be answered… though I will let readers themselves be the judge.
The first concerns consciousness and its relation to intelligence: how do we know if an intelligent machine is conscious?
The second concerns the nature of art: how do we define art, to the extent that this is even possible?
The latter is not only of concern to philosophers; it is important to everyone, not least because the art world has insistently given pride of place to works philosophically invested in the problem of art’s essence.
For me, such works have become derivative, redundant, and dull. And I think I am not alone. Many of us—including art snobs—at this point find art which poses the question “how am I art?” to be boring.
So what’s new? What’s new is that an AI-generated work was submitted to an art contest, and won (https://mymodernmet.com/ai-generated-...).
Certain artists who didn’t win complained that this was unfair. Why should an image that took little or no skill to make—or rather, to automatically generate—be eligible? And also: why should an image not made by the person who submitted it even be allowed to win?
This of course is not the first time something that wasn’t made by the person who submitted it was exhibited as an artwork. Marcel Duchamp was the first to do this, in 1917 (https://www.artsy.net/article/artsy-e...). Then again, photographers do this all the time.
Jason Allen submitted his artwork Théâtre D’opéra Spatial to the Colorado State Fair’s annual art competition, despite the fact that it was generated by the AI program Midjourney.
One key difference between Allen’s submission and, say, Duchamp’s readymade, a urinal entitled Fountain, is that the former addresses itself to the eye. Fountain addresses itself to the intellect, not the eye; in other words it doesn’t please by its intrinsic sensuousness or the artist’s skill in capturing life. It marks a turning point: instead of prompting viewers to contemplate the work’s sensuous forms or how the likeness was made (for there wasn’t any), Duchamp was here asking viewers to contemplate the meaning of art.
As for photography, a photograph can be art or not-art. Yet it is very common for photos to be “framed” and treated as art. Irrespective of an art photograph’s pleasingness, interestingness, disturbingness, it is nearly always the case—in light of how art photographers behave and how audiences evaluate art photographs—that the photographer-as-artist wishes for people to contemplate the photograph rather than treat its content as information to be understood and discarded once digested, as tends to be the case when photography is used to merely document physical reality.
Another important characteristic—which much photography and Allen’s process share—is selection. It is relevant that Allen’s AI-generated image and most photos are selected, as opposed to “created.” In the same way Allen supposedly spent hours selecting and modifying his image, a photographer selects an object or visual phenomenon to then be captured as a photograph. The artist in such instances is less a mother, so to speak, than a midwife.
The most crucial difference, however, between an AI-generated work and a photograph is that, unlike a photo, the former comes into existence as the result of combining or assembling disparate elements from a memory database in what we might call—metaphorically—the AI’s imaginary, which of course is not to say that Midjourney is a person or a conscious being. It is a machine that has been programmed to make non-repeatable images in response to a prompt.
The AI art-image is novel in this regard. Where the photograph once caused painters to worry they had been outdone by a machine in their aim of representing reality, the AI work now makes artists anxious that their crafted products of imagination will soon be obsolete. What will this mean for art?
It will surely not mean what happened once before—with painters abandoning representationalism and calling into question mimesis as art’s very basis, followed by Duchamp’s works implying that art can be anything so long as it is presented as art, in an art context.
It is conceivable that AI has put an end to the Duchampian-Warholian variety of art—that is to say, kitsch—which has dominated the art world in the two artists’ wake.
I don’t mean to say Duchamp and Warhol themselves were makers of kitsch, but that their imitators continue to produce it. Jeff Koons and Damien Hirst have enriched themselves nauseatingly by it. They and others have reiterated the riddle: “This may seem like it isn’t art… but can you really deny that it is? And what is art? Is it not like nature, everything and nothing? Is it not fundamentally the same as any item of prosaic culture?”
Wealthy collectors recognize this conceit, which now more than a century hence no longer is shocking to most of us. They recognize a signature style from a coffee table book, and conclude that it must be legitimate. The art world has become the Disneyfication of modernism: art become fashion; it is a financial bubble which will burst—and hopefully soon.
How might it end? The way we treat AI art has implied a definition of art that accords with people’s behavior with respect to artworks:
1)Art is a cultural (rather than natural) object or scene that someone (i.e. the artist) hopes will arrest the attention of others (qua active contemplation) by any means and by virtue of said cultural assemblage’s intrinsic properties. Such mere intention makes something art the majority of the time, even if the intention fails.
2)Art is a cultural object or scene not intended as art but fascinating nonetheless, provoking active contemplation by any means and by virtue of its intrinsic properties.
Thus something can be a work of art simply by virtue of its intention to be evaluated or treated as such; or simply by virtue of its reception as such.
Allen submitted his image as art because he found it arresting and by virtue of its own properties. Importantly, he thought some would feel similarly about it. All this implies art is not necessarily craft or skilled workmanship (though, I will explain, it does not exist in nature), even if they have more reliably yielded works worth contemplating.
Art is a concept unstable at its core, because it is social. The regular in the bar who repeats the same stories before an audience each night may really mean it when he says “Listen, I am not an artist, I’m just telling you how it happened to me!”
Though the audience already know the story, they wish to hear the retelling of it in order to contemplate it as a composition, in its structure. Why? It is not reality or information they are after, but the form and unfolding of his tales, and maybe his dramatic flair.
To an honest viewer, successful art is reality’s intensification. (Science by contrast is reality’s abbreviation….) Instead of attempting to intensify reality, the heirs of Duchamp’s readymade are just re-asking the already familar question What is Art? Yet by now we either believe art can be anything at all as long as it is called art, or we know that art is about trying to invite or provoke contemplation.
Perhaps when an AI begins to create images unprompted for others which it vehemently defends as art, or refutes the definition I have just put forth, we will be able to say that an artificial intelligence has passed the Turing Test and is conscious. Only something that is at once conscious and intelligent can contemplate, whether philosophically or aesthetically. We are beginning to see that raw intelligence isn’t enough to produce consciousness.
To contemplate something in nature (including one’s death), or an artwork, is to experience it for its own sake. Contemplation precedes computation, which is a means to an end. It presupposes not only a capacity for self-reflection and a concern for the phenomena of the world, but intelligence as well. At the moment AI has shown merely that it can be “intelligent,” which of course includes the ability to learn. But it has yet to prove it is actually conscious, despite recent reports of an AI expressing anxiety about being turned off (https://www.techtimes.com/articles/27...).
When an AI meaningfully contemplates its being—rather than (imitatively) saying it has an emotion—it will have demonstrated consciousness. Perhaps the best way to demonstrate the capacity for contemplation would be to take on an unsolved problem in philosophy—a problem that cannot by definition be answered by science—and argue a position that was not merely a regurgitation of an existing answer, or answers, to the problem… The definition of art, and the meaning of life, are but two important examples.
What is consciousness’s likely function? It enables us to change, to be our own creators. Consciousness makes possible reflection and contemplation, giving the capacity to adapt in revolutionary ways, to produce new or different images of what should be, beyond what currently is, to transcend instinct or habit, if we so choose. Philosophy, science, and art seem to grow out of this consciousness-dependent capacity. Artworks make or try to make us contemplate, but they also arise out of contemplation, out of dissatisfaction with what merely is, or often out of the desire to distill, intensify what is.
Which leads to the next question—what will happen with art and AI? It is difficult to know for sure. AI-generated images at the moment seem to feature (unintentionally) disquieting effects and affects that fall under the aesthetic umbrella of the “uncanny valley” (https://www.diggitmagazine.com/blog/g...). This will eventually be corrected. Audiences will soon have less and less ability to discern whether a given work (including music) was AI-generated. “Did an AI make this or not?”
And once the novelty of AI-generated art wears off—regardless whether individuals make art in the old analog way or digitally—there may be an enthusiastic return to creating works which try to invite the contemplation of their sensuous content and structure.
. . .
Innovation in the field of artificial intelligence has reached a stage where two important problems of analytic philosophy can finally be answered… though I will let readers themselves be the judge.
The first concerns consciousness and its relation to intelligence: how do we know if an intelligent machine is conscious?
The second concerns the nature of art: how do we define art, to the extent that this is even possible?
The latter is not only of concern to philosophers; it is important to everyone, not least because the art world has insistently given pride of place to works philosophically invested in the problem of art’s essence.
For me, such works have become derivative, redundant, and dull. And I think I am not alone. Many of us—including art snobs—at this point find art which poses the question “how am I art?” to be boring.
So what’s new? What’s new is that an AI-generated work was submitted to an art contest, and won (https://mymodernmet.com/ai-generated-...).
Certain artists who didn’t win complained that this was unfair. Why should an image that took little or no skill to make—or rather, to automatically generate—be eligible? And also: why should an image not made by the person who submitted it even be allowed to win?
This of course is not the first time something that wasn’t made by the person who submitted it was exhibited as an artwork. Marcel Duchamp was the first to do this, in 1917 (https://www.artsy.net/article/artsy-e...). Then again, photographers do this all the time.
Jason Allen submitted his artwork Théâtre D’opéra Spatial to the Colorado State Fair’s annual art competition, despite the fact that it was generated by the AI program Midjourney.
One key difference between Allen’s submission and, say, Duchamp’s readymade, a urinal entitled Fountain, is that the former addresses itself to the eye. Fountain addresses itself to the intellect, not the eye; in other words it doesn’t please by its intrinsic sensuousness or the artist’s skill in capturing life. It marks a turning point: instead of prompting viewers to contemplate the work’s sensuous forms or how the likeness was made (for there wasn’t any), Duchamp was here asking viewers to contemplate the meaning of art.
As for photography, a photograph can be art or not-art. Yet it is very common for photos to be “framed” and treated as art. Irrespective of an art photograph’s pleasingness, interestingness, disturbingness, it is nearly always the case—in light of how art photographers behave and how audiences evaluate art photographs—that the photographer-as-artist wishes for people to contemplate the photograph rather than treat its content as information to be understood and discarded once digested, as tends to be the case when photography is used to merely document physical reality.
Another important characteristic—which much photography and Allen’s process share—is selection. It is relevant that Allen’s AI-generated image and most photos are selected, as opposed to “created.” In the same way Allen supposedly spent hours selecting and modifying his image, a photographer selects an object or visual phenomenon to then be captured as a photograph. The artist in such instances is less a mother, so to speak, than a midwife.
The most crucial difference, however, between an AI-generated work and a photograph is that, unlike a photo, the former comes into existence as the result of combining or assembling disparate elements from a memory database in what we might call—metaphorically—the AI’s imaginary, which of course is not to say that Midjourney is a person or a conscious being. It is a machine that has been programmed to make non-repeatable images in response to a prompt.
The AI art-image is novel in this regard. Where the photograph once caused painters to worry they had been outdone by a machine in their aim of representing reality, the AI work now makes artists anxious that their crafted products of imagination will soon be obsolete. What will this mean for art?
It will surely not mean what happened once before—with painters abandoning representationalism and calling into question mimesis as art’s very basis, followed by Duchamp’s works implying that art can be anything so long as it is presented as art, in an art context.
It is conceivable that AI has put an end to the Duchampian-Warholian variety of art—that is to say, kitsch—which has dominated the art world in the two artists’ wake.
I don’t mean to say Duchamp and Warhol themselves were makers of kitsch, but that their imitators continue to produce it. Jeff Koons and Damien Hirst have enriched themselves nauseatingly by it. They and others have reiterated the riddle: “This may seem like it isn’t art… but can you really deny that it is? And what is art? Is it not like nature, everything and nothing? Is it not fundamentally the same as any item of prosaic culture?”
Wealthy collectors recognize this conceit, which now more than a century hence no longer is shocking to most of us. They recognize a signature style from a coffee table book, and conclude that it must be legitimate. The art world has become the Disneyfication of modernism: art become fashion; it is a financial bubble which will burst—and hopefully soon.
How might it end? The way we treat AI art has implied a definition of art that accords with people’s behavior with respect to artworks:
1)Art is a cultural (rather than natural) object or scene that someone (i.e. the artist) hopes will arrest the attention of others (qua active contemplation) by any means and by virtue of said cultural assemblage’s intrinsic properties. Such mere intention makes something art the majority of the time, even if the intention fails.
2)Art is a cultural object or scene not intended as art but fascinating nonetheless, provoking active contemplation by any means and by virtue of its intrinsic properties.
Thus something can be a work of art simply by virtue of its intention to be evaluated or treated as such; or simply by virtue of its reception as such.
Allen submitted his image as art because he found it arresting and by virtue of its own properties. Importantly, he thought some would feel similarly about it. All this implies art is not necessarily craft or skilled workmanship (though, I will explain, it does not exist in nature), even if they have more reliably yielded works worth contemplating.
Art is a concept unstable at its core, because it is social. The regular in the bar who repeats the same stories before an audience each night may really mean it when he says “Listen, I am not an artist, I’m just telling you how it happened to me!”
Though the audience already know the story, they wish to hear the retelling of it in order to contemplate it as a composition, in its structure. Why? It is not reality or information they are after, but the form and unfolding of his tales, and maybe his dramatic flair.
To an honest viewer, successful art is reality’s intensification. (Science by contrast is reality’s abbreviation….) Instead of attempting to intensify reality, the heirs of Duchamp’s readymade are just re-asking the already familar question What is Art? Yet by now we either believe art can be anything at all as long as it is called art, or we know that art is about trying to invite or provoke contemplation.
Perhaps when an AI begins to create images unprompted for others which it vehemently defends as art, or refutes the definition I have just put forth, we will be able to say that an artificial intelligence has passed the Turing Test and is conscious. Only something that is at once conscious and intelligent can contemplate, whether philosophically or aesthetically. We are beginning to see that raw intelligence isn’t enough to produce consciousness.
To contemplate something in nature (including one’s death), or an artwork, is to experience it for its own sake. Contemplation precedes computation, which is a means to an end. It presupposes not only a capacity for self-reflection and a concern for the phenomena of the world, but intelligence as well. At the moment AI has shown merely that it can be “intelligent,” which of course includes the ability to learn. But it has yet to prove it is actually conscious, despite recent reports of an AI expressing anxiety about being turned off (https://www.techtimes.com/articles/27...).
When an AI meaningfully contemplates its being—rather than (imitatively) saying it has an emotion—it will have demonstrated consciousness. Perhaps the best way to demonstrate the capacity for contemplation would be to take on an unsolved problem in philosophy—a problem that cannot by definition be answered by science—and argue a position that was not merely a regurgitation of an existing answer, or answers, to the problem… The definition of art, and the meaning of life, are but two important examples.
What is consciousness’s likely function? It enables us to change, to be our own creators. Consciousness makes possible reflection and contemplation, giving the capacity to adapt in revolutionary ways, to produce new or different images of what should be, beyond what currently is, to transcend instinct or habit, if we so choose. Philosophy, science, and art seem to grow out of this consciousness-dependent capacity. Artworks make or try to make us contemplate, but they also arise out of contemplation, out of dissatisfaction with what merely is, or often out of the desire to distill, intensify what is.
Which leads to the next question—what will happen with art and AI? It is difficult to know for sure. AI-generated images at the moment seem to feature (unintentionally) disquieting effects and affects that fall under the aesthetic umbrella of the “uncanny valley” (https://www.diggitmagazine.com/blog/g...). This will eventually be corrected. Audiences will soon have less and less ability to discern whether a given work (including music) was AI-generated. “Did an AI make this or not?”
And once the novelty of AI-generated art wears off—regardless whether individuals make art in the old analog way or digitally—there may be an enthusiastic return to creating works which try to invite the contemplation of their sensuous content and structure.
. . .
No comments have been added yet.
