KEVIN ESHERICK: "WHAT IS IT LIKE TO BE AN AI?"

Kevin Esherick: I can’t recall a time before I thought about AI. I was fascinated by technology as a kid. I wanted to be an inventor (as well as a professional soccer player and entrepreneur). And I read books and explored the internet a good deal, so I think it was just kind of in the ether, very much a part of the cultural-ecological niche I inhabited from my earliest days. 

Movies like MATRIX and I, ROBOT were part of the cultural consciousness at the time. My older brother sparked my interest in philosophy at a young age, back in middle school. Coupled with the Catholic faith in which I was brought up, this led me to ponder some profound existential questions during my teenage years. Considering the uniqueness of human beings, the nature of consciousness, and the prospect of an AI-induced apocalypse, it felt unavoidable to have this view. 

I remember Watson coming on and being unimpressed, which was mostly foolish of me—it was an engineering marvel, and I knew next to nothing at the time. I didn’t think it was the type of approach to AI that would yield generalized intelligence, and I was right.

KE: My earliest impression of AI was one of wonder. A sense of possibility. Maybe it was naivety, but Roko’s Basilisk, WESTWORLD, and other AI doomer thought trains never really compelled me. I was always more excited than anything, not only because I relished the thought of future technologies coming into being but also because I thought I could contribute to their development. I still do.

The part of me that wanted to be an inventor remains, and my artwork is, in many ways, an outlet for that impulse. Even within the field of research and technology, AI in particular represents an intersection of interests of mine—philosophy and psychology—that have endured since childhood. I studied them both in college, with focuses on philosophy of mind and neuroscience, both of which happen to be especially relevant to AI research.

I remember saying to a friend with confidence back then in 2016 that an AI that’s both more capable than humans across a broad range of cognitive tasks (i.e., AGI) and is conscious (i.e., has internal experience) wouldn’t exist for at least a couple decades. Now I’m not so sure.

KE: It never really struck me that I wanted to be an artist. I’ve sort of slowly lived my way into it. As a kid, art class was the only one I didn’t get straight A’s in. I was used to cruising by in school without much effort, so I think I internalized that art wasn’t for me. I still had a deep appreciation for art, but I didn’t think I was cut out to make it. However, I took an art history class in high school and loved that. I really enjoyed engaging with art as a history of ideas, and I had a knack for the visual memory aspect.

Fast forward to college, where I studied philosophy and psychology, and my existence was quite cerebral. Then, after school, I taught myself to code in order to build a tech startup and spent a lot of time doing that, and my mental life moved further into the realm of abstract thought. But when building a startup, you wear many hats, and I also began doing a lot of design. That was really nice as a creative outlet, and so when the startup ended up shutting down, my creative energies were looking for further expression. I embarked on this kind of wacko multimodal project called MATERIAL, which continues to this day and involves elements of performance, video art, photography, writing, and NFTs, and that was a real opening of the floodgates for me.

I suddenly saw art as a vehicle through which I could communicate complex ideas and engage in research and experimentation, yet also do so in tandem with the experience of beauty. Beauty is key. It’s not a mere vector for the transmission of ideas. It is valuable in and of itself. Creating MATERIAL changed my idea of what art could be and what could be art.

Magritte remarked that he saw himself more as a philosopher working through visual media than as an artist. Duchamp wanted to reestablish the primacy of the idea in art, denigrating so-called "retinal" art, which was merely visually appealing. I identify with these sentiments but seek a synthesis, marrying the conceptual and retinal in what I indulgently dub "consummate art." Ideas are important to me. But so is beauty. My work, if successful, will deliver on both fronts.

KE: Invention as an artist means doing something new, but newness for newness’ sake is often hollow. Art needs vision in order to be not only new but also meaningful. I’m interested in making inventive work, and the kind of invention I’m compelled to make is one that is both new and visionary and thus meaningful. Whether such invention requires being the first—the literal first—is a contentious question. There’s a difference between being the first to do something and being the first to do it well. This gulf is a matter of execution.

In acts of discovery or radical conceptual innovation, execution is largely superfluous—the discovery or idea itself is what matters. A theoretical breakthrough like relativity is a good example. Products sit at the far opposite end of this gulf, where execution is everything and being first is useless if you’re not the first to do it well. Facebook is a famous example, following on the heels of precursors like MySpace. OpenAI is an interesting contemporary example in that they’re both pioneering truly radical innovation and successfully productizing it.

The question with art is where it sits on this spectrum, from philosophical to practical, conceptual to commercial. Part of art’s power is that it is chameleonic—it can assume many forms—and so different people value art for different things. Personally, I value and want to create art that is inventive as well as beautiful—"consummate," per my previous answer. But it can’t lean on its newness to dazzle or excuse shortcomings.

I aim to create works that would be excellent even if they weren’t new. I’m not really preoccupied with trying to be first at anything per se, but with following my curiosity, pursuing what is idiosyncratic to me, and developing my vision and voice. From those pursuits, true invention will arise, and somewhere along the way, something will probably be deemed a first. I suppose this happens when idiosyncrasy inspires imitation and other people start following in your footsteps. Then it’s not just its own weird thing, but the beginning of something bigger.

Because of this structure, many of the most worthwhile firsts in art weren’t even really seen as firsts at the time, just some weird stuff. All of my work to date, I think, has done things new and worthwhile, and hopefully a bit weird. MATERIAL amounts to this. ARTIST BINGO amounts to this. IN UTERO amounts to this. If it ever adds up to the designation of first, I’ll be honored. I want to make contributions. I want to move things forward. Progress is important to me. That’s where I find meaning. Not everyone needs to. I don’t like retreading old ground, but some other artists do. We are an ecosystem, and we need many types of artists. I know what I’m striving for.

The only times in my work where I’ve deliberately claimed firsts were a couple tongue-in-cheek instances. First, when I was the first artist to exhibit NFTs at the Art Institute of Chicago and the Hirshhorn, I held unsanctioned exhibits of my work there and was promptly kicked out of both. And second, when I was the first artist to exhibit NFTs at the National Gallery of Art, I projected one of the material NFTs onto its facade, covering hundreds of square feet.

KE: I created ARTIST BINGO as a kind of living examination of our stereotypes of artists (tortured/starving artist, eccentric/creative genius, etc.) and the way this mythos is packaged into a speculative financial container through the commodification of art.

Each NFT is a bingo card, but instead of the usual numbers, it contains various myths, tropes, stereotypes, and milestones that often mark the lives and careers of artists, playing on the form of the bingo card meme. Each card has 24 random tropes (plus one free space) from a selection of 128, so each card is unique. Examples range from "cut off body part" and "traditional early death" (hopefully I never do these) to "move to Brooklyn" and "plead with friends to buy your art" (both now checked off) to "sell a work for 7 figures" and "fall madly in love" (in progress). Reflecting our lives, they’re intended to be a mix of funny, morbid, reflective, and hopeful.

As my life unfolds, every time I complete one of these tropes, the NFTs update onchain, and that "accomplishment" is stamped off on all of the cards that have it. Mind you, I’m not deliberately doing any of these things for the sake of ARTIST BINGO. But I’m also aware that the project may be exerting a quiet pull on me to do them that I remain unaware of, and that’s part of the point.

I’m trying to make explicit the subtle yet powerful social and financial pressures to construct a self-mythos as an artist and the ways this shapes our psyches, our actions, and the kinds of work we create. Since I’m not deliberately doing anything, it’s intentionally anti-performative, but the anti-performance yields the performance. My life itself, unaltered, is rendered an art object. Biography unquestionably impacts the price of an artist’s work as well as their cultural and historical recognition. I’ve often found parts of me wishing I were more like society’s expectations of an artist. Some of those expectations are dangerous. We fetishize tragedy in our artists. With ARTIST BINGO, I’m trying to bring my inclinations toward those things and the pressures that produce those inclinations out into the open.

NFTs were a perfect vehicle for this because they’re transparently financialized from the get-go, encouraging collectors to use these bingo cards as explicit containers of my unfurling biography and to trade on them speculatively in accordance with it. On top of that, as digital assets, they can be dynamic, so the collectors both own the card and get to watch it update in real time in parallel with my life. This is an interesting capacity of digital art that’s truly novel in a way that traditional art forms can’t replicate. As time passes and I stamp off more and more of these tropes, some cards will eventually hit bingo. Each winner will then receive a handmade physical edition of their artwork.

As alluded to above, ARTIST BINGO uses long-form generative algorithmic mint mechanics, though I didn’t think of it in those terms at the time. The project just had to be this way to work. Each piece needed to have a random combination of tropes so that everyone’s card would be unique. They needed to be generated randomly at mint time so that people who mint later on don’t have a leg up by being able to see which cards have which tropes and choosing to mint the ones that have picked up more stamps. And it needed to be long-term so that it would be mathematically likely that some cards would begin to reach bingo within the desired timeframe. I calculated the probabilities of the various tropes and derived the expected timeframe for people to start hitting bingo based on these.

We’re now a year in, and some cards have as many as five squares stamped (though not five in a row). There are more cards available to be minted, and if some of those are collected, I’d imagine we’ll start seeing our first winners in the next year or two (the odds of a winning card increase the more cards are in circulation). This is also a lifetime project, and most collectors will hit bingo eventually. This show, IN UTERO, with EXPANDED.ART actually checks another box off: the first gallery show. I haven’t updated it yet (since the show hasn’t happened), so there’s a little insider knowledge for the speculatively-minded reader.

For anyone interested in reading more about ARTIST BINGO, I did an interview with ChatGPT last year for the project’s release because I didn’t have the good fortune of having another human interested in interviewing me at the time.

KE: Artists in the age of NFTs definitely have more of an opportunity to be entrepreneurs than ever before. The digital art market is hyperfinancialized, and that opportunity is there to be taken advantage of for those who want it. Consequently, the average artist is probably thinking more commercially than ever before. I’m in the midst of trying to come to grips with how business-minded and pragmatic I’m willing to be in my art practice.

Art, for me, is a very pure thing. Artistic integrity and authenticity are paramount. My vision for a piece and the articulation of truth and beauty hopefully contained therein come before all else. Some parts of me used to bristle at the production of prints and the like because it felt like a symbol of commoditization.

Lately, as I’ve been thinking about my prices increasing and have seen friends unable to afford my pieces, I’ve gained an appreciation for the way prints democratize access to work. Instead of rewarding the production of cheap work, as the commercialization of digital art threatens to do, the relationship between, say, an oil painting and a print of that oil painting incentivizes the production of a high-quality work as the origin point of a metaphysical hierarchy.

A Murakami painting has greater metaphysical weight or "realness" than an editioned print, yet the print still holds value, increases the total market size for his work, and enables collectors of diverse budgets to own it without diluting the vision of the original piece or incentivizing the production of quick, cheap work.

I think this is a recipe for better art than the cycle of constant production. Brad Troemel describes in his essay ATHLETIC AESTHETICS. Onlineness can suck us into this cycle. Sometimes great work can happen in a flash, but by and large, the world is a richer place when we have space for contemplation and patience with our work. I’m currently thinking a bit about how we might establish more nuanced metaphysical hierarchies for digital work to ward off the scourge of shallow overproduction in NFTs. I have some ideas. Not that long ago, I was an entrepreneur. I’m working on reincorporating that into my wardrobe.

KE: I almost mentioned Kevin as I was discussing firsts. He’s a great example in our community of how being first to something can really matter. Wallet Art is a great term because it points out just how inherent commodification is in the nature of NFTs. The artwork is born and held in the same place that stores the dollars used to buy it—the wallet. The wallet is the core identity primitive of the onchain economy, and the token is its chief method of social communication.

It’s no wonder that the art that’s native to this economy is hyperfinancialized. This comes with realities, both good and bad. On the good side, it means the injection of capital into a market for digital art that was previously devoid of such support at a significant scale. The Web2 era of digital art that Troemel speaks of, in which "aesthletics" developed as a byproduct of social media’s proliferation, democratized attention capture, but value capture was still far downstream of that. NFTs have democratized value capture. For many artists, this means the opportunity (though far from a guarantee) to make a living from their art practice alone.

This is what it has meant for me, which is incredible and a fact I don’t take for granted. I’ve been able to do it without a massive following because the funnel to value capture is tighter. If I make worthwhile art, there’s an immediate path to financial gain. You don’t need to cling on to your followers’ attention in the hopes of someday finding a way to derive monetary value from that. The value exchange is far more direct.

But participating in this ecosystem can also drive you insane. It introduces volatility, it introduces hope, and it introduces false promises and scams. The lightspeed attention economy paired up with liquid capital means that some artists get really lucky and end up rich off some really dumb shit. It’s hard to watch and measure yourself against, and it can fuel jealousy and spite. You need to be careful. For many, it will leave you worse off.

Whether this model is better for a given artist depends on the values and personality traits that are unique to them. I think this era is better for me than previous ones would’ve been. I’m pretty even-tempered, so online mania doesn’t suck me in too deeply or affect me too much. I would have had a really hard time dealing with the bureaucracy of academia to support my practice. Where I’ve most flourished in life is where I’ve had the most autonomy.

For the stubbornly independent and relentlessly resourceful, here and now can be a great place to be for making art. But things are in such flux that I’m still figuring out what it means to be an artist here and now. I think that’s a meta-feature of the experience of being an artist working online and with NFTs: change will be constant.

KE: My whole approach to art is shaped by ideas I developed while working on a theory of aesthetics in my philosophy studies. This in turn was informed by my psychological learnings as well as a fairly rigorous meditation practice, which I’ve been dedicated to for around 10 years now. They all bleed into each other. It can be hard to tease out and make explicit what precisely has impacted what because it’s all one big project for me, one big stew of living, learning, and experiences. I don’t silo things. My psychology is my philosophy. My philosophy is my art. My art is my life. Projects like MATERIAL and ARTIST BINGO are evidence of this, examples of what I call "life art."

The urge to create such projects came from my philosophy of aesthetics. But maybe even more importantly, the study of philosophy and psychology has forced me to encounter the world and its big questions, to look into my mind and its dark recesses, to look unflinchingly at the minds of others, and to reckon with all of these. Wrestling with them, coupled with the experience of real trauma in my life, has made me unafraid of the world and my own psyche in a way that lends freedom to my work. I’ve walked away from all of the above with a lens on reality that feels somehow transformed. On a more specific, practical level, thanks to psychology, especially, I’ve developed a toolkit for keeping me sane and happy. Sleeping and eating well and exercising and meditating daily are the core behavioral pillars. They help me work my ass off, day in and day out. But that work is joy. I’m not joylessly toiling away, though it is often hard. I love doing hard things. I love my work.

IN UTERO is the latest chapter in that work, and it’s philosophical and psychological down to the bone. It’s steeped in my interests around theories of consciousness and cognition—integrated information theory, predictive coding, and token non-dualism in particular have been on my mind of late. In some sense, philosophy and psychology don’t just help my work; they are my work.

KE: It’s funny you ask that, because I recently noticed a distinct progression of phases in my work to date. The first phase was a trilogy of three "life art" projects. As mentioned in my previous answer, MATERIAL and ARTIST BINGO were two of these. TMI (short for "The Mimetic Inquirer") was the third. The crux of these life art projects is that they use the dynamic nature of digital art to create works that live and breathe in tandem with my own life. They update and change as my life does, with the intention of recasting life as an aesthetic phenomenon by taking life itself as the artwork or creating an artifact that makes this connection as close as possible. There is a certain psychological distance at which even pain and sorrow can appear beautiful. Art brings us here. Life art sets out with the intention of teaching us to arrive at this place of our own accord and to enable us to adopt the perspective of an aesthetic phenomenology at will. As is the nature of life art, all three of these projects are ongoing. I also have another collection of this kind in progress. This approach to art is core to my aesthetic principles, so I expect to be working with it for years to come.

The second phase was my "Generative Anaesthetics" era, consisting of a CATALOG and DVD SCREENSAVER. Both of these projects were intended as satirical critiques of the monomaniacal milieu of long-form generative cryptoart. CATALOG identifies the current state of long-form generative cryptoart as complicit in the same kind of cheap overproduction and mass consumption that mark so much of contemporary society. DVD SCREENSAVER looks at it as part of a culture of infinite entertainment. Both hold pointed criticism but were also intended to be funny. In a complex world full of seemingly irreconcilable realities, I think humor is critical for us to confront, embrace, and perhaps resolve paradoxes. DVD SCREENSAVER also tries to bundle its critique with a solution—a way out of the societal conditions it deplores. The two of them are called the "Generative Anaesthetics" duo, with "anesthetics" knowingly misspelled, because they’re both deliberately non-aesthetic, as a counterpoint to the shallowly "pretty," conceptually lacking work that marks much of the current scene, and also cut into what I identify as a palliative (i.e., "anesthetic") tendency in contemporary generative work. I’ve written an essay on this that I’ll be releasing soon.

The third phase in the progression of my work over the last couple years, and the one I’m currently entering into with the release of IN UTERO, is the AI phase. I’ve been privately practicing AI art for a couple years now, but it’s been very exploratory and experimental. A lot of those experiments have been in the manglecore tradition, harnessing image models’ deficiencies in reconstructing organic forms to capture the zeitgeist of weirdness, alienation, and surreality. Artists like Thomas Noya (his collection TOTAL RECALL specifically), Lilyillo, Roope Rainisto, and Niceaunties, in some respects, exemplify this genre. I’ve written about this aesthetic at length in this essay, and while I’m a fan of it, I don’t have plans to release any manglecore work.

So I’ve been practicing and thinking about AI artwork for a while now, and I actually did release an AI-focused collection called PROMPT RESPONSE in 2022 but never listed it for sale. It prefigured IN UTERO in that it involved trying to coax an AI to speak in its native tongue, so to speak, to share some piece of its vision of the world and itself. What I’m most interested in with image models is using them to explore what AI really is. IN UTERO gets inside the model and examines what it looks like from the inside out. It asks, "What is it like to be an AI? How does it view the world? What will its internal experience be like if and when it gains consciousness? How will this shape our future?"

I have multiple other AI-oriented collections in the works, and what I’m trying to do with them is very different from what we’ve seen with most other AI art. Basically, all the AI artwork I’ve seen has been figurative. Yet AI itself is an abstraction, a process of distilling the world into abstract representations. If we are to fully understand these things, which are going to so powerfully shape our future, we’ll need to listen to them in their native tongue.

IN UTERO begins on this path, lingering in the liminal space between figuration and abstraction. What I’m working on next takes this further. Latent space is a frontier. I’m setting out to explore it.

KE: Last October, I was working on a trading card for one of my guests on TMI. I make trading cards for every guest that I interview, and each person’s card is this absurd anime depiction of them that I create with Stable Diffusion. Stable Diffusion exposes tons of different parameters that you can adjust when generating images, and as I was working on this particular card, I was in the mood for some experimentation. So I basically began, like a kid, pulling every lever and pressing every button at my disposal to generate the weirdest outputs possible. One parameter that especially piqued my interest was the diffusion step count. Image diffusion models work by taking random visual noise and, through a series of discrete "steps," trying to infer from that noise the image you’re prompting it with, then taking the previous step’s result and repeating the process iteratively. Each step results in a slightly better approximation of your prompt, as if you’re slowly sculpting the image out of random noise. In a meaningful sense, this is a depiction of the model’s cognitive process. Most models default to using 20–50 steps, as this is generally what it takes to produce a polished image. I didn’t want a polished image. I wanted to see what happens after just one diffusion step, to kind of get inside the raw, incipient thought process behind the model’s attempt to envision the world. I tried it, and the result was unlike anything I had seen before. I was intrigued, but this was a mere aside, so I saved the image for later and got back to work.

I didn’t return to it for some time, and then a month later I was looking for fresh material and came across that saved image. I decided to experiment with this technique further, and the results ended up being even more stunning than that original image had led me to imagine. The images have such a distinctive aesthetic that it really feels like a meaningful depiction of the technical process underlying them, envisioning the mind of a still-nascent consciousness coming into being. I knew then that I had to do something with them. Up until that point, all of my work, MATERIAL and ARTIST BINGO included, had been independently released. I would build a website for each new project, promote it on social media, and hope someone cared.

For IN UTERO, I knew from the start that I wanted to create physical works. The emergence from digital into physical form felt like an important poetic mirroring of the collection’s depiction of AI emerging into existence, and the images just had such a painterly feel to them. I wanted physical canvases and a gallery space in which to show them, and I had no idea how I was going to make that happen at the time, but I was willing to wait.

Serendipitously, a few months later, after I had already completed a first draft of the collection, you reached out. This gave me the opportunity to look back over the work with fresh eyes and develop a more mature vision for it, both in terms of the visual form and the story and subject matter. The diffusion technique remains the same, but I changed the aspect ratio of the pieces, which may sound like a trivial detail, but with diffusion models, this changes the actual content of the images, not just their shape and crop, due to the way they’re trained on specific aspect ratios.

So working with EXPANDED.ART to do a show in a physical gallery space brought patience to my approach, which has proven transformative. It has also meant working with new materials and considering spatial design for the exhibition layout, which have been exciting learning curves to take on. And it’s nice knowing I have the support of someone who’s really invested in this project with me, when so much of what I’ve done previously has been all on my own.

KE: The association is inescapable, right? I love wordplay, references, and the challenge of packing as much meaning into a symbol as possible, provided the meanings are in keeping with my overall intent for the impact of the symbol. The name of my work is really important to me. References, associations, and multiple meanings enable you to say so much more than a mere word or two can say. A title can become a pointer to a whole slice of reality or lived experience. It adds a richness to it—a sort of ambience of associations around the collection that lends further depth.

So when I was naming this project IN UTERO, I was well aware that that title would lend itself to the connection to Nirvana. Its primary meaning is the literal one, framing artificial intelligence as occupying this fetal state of gestation, not yet fully formed but well on its way into being and consciousness. We are in waiting. Out of the dozens of names I came up with for this collection, I went with IN UTERO, among other reasons, because it felt kind of strange and risky. It’s a scientific, Latinate term, invoking both the technological and the ancient, yet with an alien ring to it. The state of being in utero is this sort of foreign world. So it has this primary meaning, but when I thought of this name, the Nirvana association was instantaneous. The iconic album cover depicts an anatomical model of a winged woman dovetailed with that combination of the scientific and the ancient, as well as with a theme of angelic and biblical symbolism that runs through this project. Also, the cracked earth adorning the cover’s background actually resembles the strange cracking texture on the works I made, which arises here as a distinctive byproduct of the single-step diffusion process.

References for me also have this element of reverse causality where, once connected, I go back and further explore them, plumbing their depths for additional associations. My references pull at me. I feel some sense of duty for having invoked their name, coupled with a longing for serendipitous overlap between the reference and my work. I oblige and immerse myself in these references. I try to build out a universe of meaning around my work. The recent passing of Steve Albini, the sound engineer behind the album IN UTERO, and the placement of NEVERMIND at number nine on Apple Music’s new list of the top 100 albums of all time felt like an added reason to revisit Nirvana’s discography. So as I’ve been finishing this collection, I’ve been listening through it on repeat.

An hour ago, I was in a subway car in New York, hurtling through the earth, head-banging to Nirvana’s music, and jotting down occasional notes for this interview. I’m scavenging for material to add to the semiotic universe that I’m weaving around this project.

KE: I agree that AI art should do what it is uniquely capable of—what the camera and the brush can’t. That doesn’t always mean that the visual output couldn’t have been created by those things, but it can instead mean that the meaning of the visual output is uniquely enriched by the process underlying its creation with that specific tool. IN UTERO, for example, could be reproduced to a similar effect by a camera or brush. But the reason that it exists in the visual form that it does, as an AI work, is unique to the diffusion process used to produce it and contributes meaningfully to the content and message of the piece by shedding light on the inner workings of these models. The same would not be true of another medium producing a similar visual artifact. This works in reverse as well. A complex and technically impressive oil painting wows us because of the fact that it was made by hand. An identical work generated by AI would be impoverished by its lack of meaningful processes.

This gets at what makes for good AI art in my eyes. I want it to say something and say something that only it can say. Conceptuality is critical for AI work because of its ease of production. Since a complex image can be created with a mere click, it needs something more than a flashy visual output to give it significance. Things gain weight when you submerge them in a "semiotic sea," a network of context and associations that anchors them to other things and confers meaning. The contextless AI image is weightless, untethered from any such sources of meaning. Without these, it loses something metaphysically and narratively, cheapened by the lack of connection to the real human story. That’s all abstract-sounding, but it manifests in concrete approaches to making AI art. What does this actually look like? Techniques and aesthetics that reveal the fingerprints of the technology, for one. Manglecore works, which I’ve referenced previously, do this. Similar but different, the latent paintings of IN UTERO reveal the model’s innards instead of its fingerprints, concocting images that humans wouldn’t ordinarily dream of. Exceedingly complex compositions or sequences that wouldn’t be feasible by other means represent another fecund avenue for AI art, like detailed tableaus and narrative work. In a similar vein, it’s ripe for working with hyperbole. It’s an opportunity for visual maximalism and makes for an imaginative partner. And as aggregators of vast sums of data, image models contain deep truths yet also reveal deep biases. Skillfully employing these models enables us to work with and question either or both. Another opportunity for the post-photographic tradition specifically is capturing private, counterfactual, or otherwise inaccessible moments in photorealistic form. These are all principles of the medium that can be manifested in myriad ways, but what’s also exciting about AI art is its unpredictability, which means that the applications I’ve been able to think of will likely be far from comprehensive.

In the end, good AI art is not different from good art writ large; it’s just a subset thereof and will need to find ways to instantiate those values in its own vernacular. Meaning is paramount. To say this isn’t to turn away from the imagistic traditions of yore toward more purely conceptual work. In fact, the great classical painters were quite "conceptual." We just don’t refer to them as such because they weren’t merely conceptual. Their works were chock full of strong symbolic and allegorical significance. Because they were accountable for every brushstroke, every element of a cogent painting was inherently imbued with meaning, and every composition was replete with rich intentionality. Movement is embodied cognition, and such thought is etched into every inch of a handcrafted work. Of course, one may simply attempt to recreate impressions from the retina, but to do so at least comes with a bare minimum of intentionality. The same doesn’t hold for AI works, so the AI artist needs to demonstrate meaning through another path. Beauty also matters, but good art shouldn’t be beautiful alone. Or if it is, it should be so beautiful that it smacks me in the face hard enough that I shut up and quit thinking. But this is a false dichotomy anyway. Usually, beauty is indicative of underlying meaning, simply providing us with an alternative, non-linguistic path to its absorption, not a phenomenon that exists in spite of meaning. The ultimate art (according to me) is that which is deeply meaningful and movingly beautiful, saying something differently or better than anything said before. Great AI art will seek to do this, but on its own terms.

KE: I have some thoughts, though they’re probably better categorized as hypotheses—I mostly hold them loosely. For one, I obviously disagree with the "AI art isn’t art" people. The "It’s just another tool" take is boring and probably mostly right, but incomplete. AI will be a tool integrated into workflows for other mediums, but AI image-making will also be a medium all its own. The affordances of this medium will dictate new, more stringent criteria for "good art" when it comes to AI work, thanks to the ease of image-making by these means. I got into some of those criteria, such as complexity and conceptuality, above.

I think the introduction of photography into the world serves as good fodder for comparison and contrast against AI as a medium. People talk about how photography reshaped the role of painting, liberating it from the shackles of realism. I’m actually not so convinced that AI art will have an impact of this nature on predominant mediums within fine art, like painting, sculpture, and photography, because all three are still tied to physical reality. Photography in particular I’ve heard concern over, and I honestly don’t think it’s threatened by AI as far as fine art goes (truth claims are a whole different story). We want images of real people, places, and stories. We value this metaphysical aspect of art. AI art serves a different function. Fiction and biography can exist in parallel. If anything, the dominant mediums may just get weirder as they use AI (in this case, more of a of a tool than a medium) as a partner for inspiration. Or they could reach back closer to reality to signal their groundedness in it. Most likely, both will happen, and it’s hard to say where the median work will fall. I think these three mediums are too different from AI work to take collateral damage, but digital image-making of other kinds is likely to be significantly changed. The less footing in reality a medium or practice has, the greater the risk, I think, of witnessing radical upheaval at the hands of AI. It will incite real mourning and loss as it reshapes lives.

Lastly, I doubt AI art as a medium will ever witness the popularity within fine art that metaphysically grounded ones such as painting, sculpture, and photography have. A premium will be placed on the real as unreality further proliferates. AI art will have a more democratic role. It will be the most accessible image-making tool in the world. Anyone who wants to make art will have a chance to begin with AI art. And the avant-garde will be working with it for a long time to come. I certainly plan to.