News center
Superior after-sales care

Matthew Ritchie with Jason Rosenfeld

Jul 15, 2023

Matthew Ritchie's show, A Garden in the Machine, is at James Cohan at 48 Walker Street through October 15. It includes two series of paintings made in the past year, a suite of ten related drawings, each titled Leaves, a large sculpture, and a film. The artist's major career survey, A Garden in the Flood, curated by Mark Scala, will open at the Frist Art Museum, Nashville, Tennessee, on November 11. It will also include a collaboration with the composer Hanna Benn and the Fisk Jubilee Singers, with direction from their recently deceased leader, Dr. Paul T. Kwami. This is Ritchie's first solo show at the gallery.

Jason Rosenfeld (Rail): Can you walk us through what's going on in these new paintings, and the process, and the question of Artificial Intelligence-aided art, and subverting AI?

Matthew Ritchie: These were an offshoot of working through COVID. I started the process at MIT three years ago where I was an artist-in-residence at the Center for Art, Science & Technology right before and during COVID. I was working with Dr. Sarah Schwettmann, a programmer, researcher, and cognitive scientist. She got invited to participate in a collaboration with MIT, Microsoft, and the Met. The Met came in and said, "So we’re releasing our collected images to the world for free use, and we want you guys to figure out what that should do, and Microsoft is going to provide the computational power." Later, the Met did a few events called digital salons, but it was at the Hackathon I first saw a kind of machine learning, called generative adversarial networks, or GANs. One of the programs in the GAN is called the generator, and one is called the discriminator. Based on a database, the generator makes millions and millions of potential images, and the discriminator chooses from them. They’re adversarial to each other, but still trying to create, or converge, on an entirely new image based on whatever data they've been given.

Rail: They don't like each other. But the discriminator gets the last laugh?

Ritchie: Yes, the discriminator gets the last laugh, so tragically true of our society. [Laughter] A new kind of GAN, called StyleGAN had just been made available in 2019, and a programmer there called Mark Hamilton ran the entire collection of the Met through it and it just made this kind of captivating nonsense. Sarah called it a "blurryvid." Because art doesn't add up like faces. If you give it 500,000 works of art, the GAN cannot make any new works of art. The software tries to sort of solve the art. And it can't, so instead you saw this thing morphing between what may be a necklace, maybe now it's a coin, now it's a painting. Wait, it's a hat! It's a vessel! It's a piece of chainmail! I felt for the first time in my life that I glimpsed a visual equivalent to what we think machine intelligence might look like. It's not that, but I felt it was the first time I could see the mind of a machine.

Rail: GAN was not developed for imagery?

Ritchie: Not originally, but StyleGAN was.

Rail: So, the original application was for what?

Ritchie: I think its initial use was to try and understand whether networks can learn to train themselves, but of course learning depends on categorizing what to learn, the subject matter in some way. So StyleGAN was trained on human faces and learned how to make more faces. The idea was, if you give them a million faces, they can make you a million more faces.

Rail: It synthesizes material and produces something, I guess you could say original.

Ritchie: It's an averaged-out version of all the things that are in there. A website called is an early example. It makes a new face and a new face and a new face, and they all look like real people. It's crazy. Now ads you might see on the subway—that won't be a real person. That won't be a photoshoot. They don't need that anymore, which is sort of scary. That's what I knew GANs as. And of course, what people were really doing with them was making celebrity mashups. So when I saw them applied to art at MIT, they didn't work—but there was a fraction of a second when I thought, "That. There. It happened in front of me. I saw it." And then it kind of dissolved into this fog. I got the programmer to give me the video that he made of that formation of images and for a year I tried to figure out how to make it work, until Sarah set me up with the same code. It's all openly available and free on GitHub but there are also now pay platforms as well. She sort of hacked together a platform for me, which is really what enabled me to make Caudex (2022), a machine learning film in the back of the show. I've obviously made a lot of animations and films as part of performances, so I always liked the idea that you're not really making a film. It's more like machinima, a procedural animation. You’re referencing a block of code, rather than a narrative. But with the GAN you can train the discriminator. You can say, "I like this," and choose what it likes. Then the two of them work hard together. Especially during the internal exile of COVID, I began to feel I was having a conversation with this thing, because it would make all these beautiful efforts to please me, most of which were failing. So, it was sort of like painting or teaching in art school. [Laughter]

Rail: Did you delineate the scale or the orientation?

Ritchie: It makes everything into a square because it's just seeing pixels. You feed it a million pixels, it will give you every arrangement of those pixels. That's all it's doing. It's not thinking. The discrimination is that once you start to identify preferences, it will try and give you more of the thing that you’ve said you like. You like more of these swirly things? I’ll give you more swirls.

Rail: Like social media. It's reading your preferences.

Ritchie: Exactly. What makes it different from the newer diffusion AI, which is sweeping the nation now, is the GAN doesn't use any text. In one way there's no overt implicit textual bias. It's just the images.

Rail: Because the material that ambles through the generator has not been catalogued with a title. It can only work visually. What about this guy in the Times who won this three hundred dollar contest at the Colorado State Fair, who made this appalling picture based on AI diffusion and text?

Ritchie: Appalling to some, brilliant to others.

Rail: I looked closely at it. [Laughter] And the fault is not with that guy, it's with the competition judges who didn't know that it was AI and picked it anyway thinking it was great.

Ritchie: They wrote something about it, which was like, "masterful." It's basically Dune.

Rail: It's Dune meets Gustave Moreau, in some weird way.

Ritchie: Now it's already sounding good again.

Rail: I guess so, until you look at it. So that's not what we're seeing in these paintings. It's something quite different.

Ritchie: That's where the GANs rather than the diffusion models, which are text based, differ. You must know what to ask for, which is a bit like you've already been trained. Whereas the GANs space is untrained. It's up to you.

Rail: It doesn't know going in, and it never will.

Ritchie: It doesn't know anything. And it never will. It's an enthusiastic idiot, much like myself. [Laughter] I noticed that when the program ran through the Met collection the moments that it began to become beautiful were when it looked at the drawings. Something is lost when it's built on a photo library, because it's quite good at making things photorealistic. Since drawing doesn't have all the reference points, like shading, the GAN can't really assemble them. And so, it frees it. And you start to see what it's really doing, which is similar to human cognition. It's looking for edges and eyes and forms.

Rail: Because it doesn't know a tree is a tree.

Ritchie: It doesn't know a tree is a tree. When the scientists working at MIT started to use it they noticed it would find things in pictures, such as reflections, but they didn't really understand how it found them. It would automatically start generating a reflection of a window and a door. And even these people, who know it's just code, get convinced that it's sort of thinking. But no, it's just numbers in a big pile.

Rail: It's seeing in a different way.

Ritchie: Yes, much like nature. It's both the history of nature, which is atoms being assembled into every possible random form. And then the history of art is whatever anyone made art of. A rope. A hat. That was what was interesting in seeing the Met's diverse collection through time. There are so many things like masks. The program loves masks because it thinks they’re faces. You realize how much our work as artists is about a kind of masking, or the presentation of a pseudo face.

Rail: So, the GAN has James Ensor posters in its office!

Ritchie: It loves James Ensor. Then some other beautiful mistakes happened. Because it can't think. And the resulting image is at a super small scale. It's just a 150 pixel jpeg. Here I’m pointing to an area in a lower layer of the painting Generator (2022). That's the blown-up jpeg. And it's just this gray. I was so disappointed when I first printed them out.

Rail: You start with the scaled canvas, which in the "Branches" series is 66 by 77 inches. And then you use a machine printer to print the blown-up jpeg from the GAN onto the canvas, this image derived from this sort of database.

Ritchie: A classic technique of the twentieth century, some sort of printing underneath. Picabia, Polke, Julie Mehretu, Rauschenberg. I’m listing all my art heroes. Max Ernst. But again, the beauty of this process is you keep going back to originals. As in the traditional Renaissance method of pouncing charcoal using cartoons to sketch out the main image on the panel or wall. What you get is this kind of blurry, faded out form. It's not really the painting at all.

Rail: It's a transcription, to a degree of the generated image.

Ritchie: But fuzzy. And then you’re on your own. It gives you the drawing, the most general sense of the drawn structure. But with the GANs, there really isn't a drawing there to begin with. It's not like anatomy. It's just a bunch of weird lines. I always found halfway through these paintings that I was painting on top of my own lines, thinking they were the GANs lines, so it becomes just the classic iterative process of an artist, alone in the studio, just reacting to their imagination. The whole thing is just a projection. There's definitely no artificial intelligence. I’m not sure there's even an artist intelligence.

Rail: Then on top of that, you work freehand. And with a certain limited palette. There are bursts of various color here and there, but mostly the palette is earth tones, a little bit of emerald, some blue. That was a decision you made consciously for this suite of paintings?

Ritchie: I work pretty intuitively, having done all this pretend thinking. [Laughter] And I felt all these artists while I was making these pictures, or weird artists that I never look at—I was convinced at one point that this was Lucian Freud's color palette, it was all going to be brown. It didn't work. Wasn't right. The color palette is sown from my own work, all these rich reds. It's very medieval. Sort of Thomas Aquinas color theory. It's like stained glass or tapestries. All those old Renaissance colorways. And a kind of sense that the colors all have autonomous meanings. Which of course they don't.

Rail: The paintings don't resemble, but they recall. They have, in a general sense, some sort of hovering forms in front of a bluish background. Sometimes you can read a little bit of a landscape element in the lower section, but maybe that's just what we do in looking at pictures, giving them weight.

Ritchie: I added all the color. Your sense of what is present in the picture is to do with me finding all those things. But while I was making them, I was discovering all those forms. I wouldn't even say they were in the GAN print, now that I sort of realized I was kind of using that. In a way it's like automatic writing, or exquisite corpses, all those surrealist processes, which were a way to turn it into a biomorphic machine. At the same time, it feels like a body. To me it is interesting that I don't know what they are, and I painted them. I kept thinking of all these artists like Wilfredo Lam, Paul Nash, Graham Sutherland, and Eduardo Paolozzi, who were all in that kind of interwar moment when the body and the machine had literally been torn apart in front of them. They’d seen it. And then they tried to come back. All these great artists and characters. And what they saw was the beautiful, Edwardian world, a fairy tale organized history where we've got all the orders of ornament and all this stuff. They watched that in Europe just be torn apart. And that seemed very timely and climate change-y and, hey, here we are.

Rail: The importance of the world wars is obviously massive, but we shouldn't denigrate what we lived through. When you live through 9/11, and then through the pandemic. It impacts you, and I think you can see that in a lot of artists who are working right now.

Ritchie: It's certainly like that disassembly that was part of the internet, that has happened to our consciousness, too. I made a lot of paintings about 9/11, because I was here when it happened, and took that very much to heart. And the resulting wars, the inversion of morality that happened around that, where in order to preserve democracy, we must become undemocratic, in order to be free, we must live in a prison. All that kind of insanity. The kind of anti-language. And having grown up with Orwell, in England there are so many precedents of this, all about similar movements in the pre-war period, and post war, and then the whole magical reaction to that in the fifties, in writers like John Fowles. I've always felt living in America that it's a culture with a medieval philosophy at its heart, its theology is medieval. Its system of government is enlightened, but it's very Manichaean, light versus dark, and you kind of hoped, much like Europe, it was getting over some of that, but oh, no. That did not happen.

Rail: It's come back with a vengeance.

Ritchie: It took Europe a thousand years of war to just kind of get tired of it. But hey, it's back in the Ukraine. Not only is the apocalypse happening in the world, but it's also happening in our brains now, thanks to the internet, and our bodies.

Rail: There's interlacing but alienation, and the planet's at risk. I think you get that sense looking at these. And talking about medieval art, Yew of the Wood (2022) has a centrality, and a sort of buoyancy of that central, rising round form, which makes me think of Last Judgment paintings, or religious works which relate to a certain kind of iconography. But a lot of this is just unconscious, just happening as you're working?

Ritchie: Well, these paintings in the "Branches" series were fed medieval apocalypses mixed with computer generated images of the large sculpture in the show, Shadow Drawing (2022). The GAN is trying to get the vibe of two incompatible things.

Rail: The sculpture was surprising. I didn't expect to see it when I walked in here, but it makes total sense visually.

Ritchie: It's a half-size version of one opening next month at the University of North Texas, which is called Shadow Garden and is set in a garden in the shadow of an enormous oak tree. That's also where the soundtrack for the film in the back, Caudex, was performed as an opera called Infinite Movement, with extraordinary music by Shara Nova, and a libretto I generated using AI and Milton's Paradise Lost, which was also to do with the iteration of time and narrative in space. This technique with the GANs was a way to think about history, but not in a linear form. And that didn't seem like something that was present when I painted the "Branches" pictures, they were all right next to each other in my studio. It was one wall.

Rail: You were working on them simultaneously?

Ritchie: Yes. It was like one giant painting. And it was not something that you could show. It was just overwhelming, the density. Because they’re somehow deep, where a lot of my earlier work was kind of wide.

Rail: It's interesting to think about them like a Renaissance fresco cycle, such as in Padua, and seeing them as broadly time based, not linear, not telling a dedicated narrative, but relating to one another. In the way that Giotto was thinking in the Arena Chapel in terms of layout using quadrants and background elements in those works that jumped diagonally or horizontally across frames and in and out of the timeline of the holy family scenes. He also harmonized them coloristically across the walls, which I guess is what I was kind of noticing with your bluish backgrounds. It's not that deep blue of Giotto, but it's a similar sort of idea. And that makes more sense to me than seeing them here as a totality.

Ritchie: It's like being a modern person, we can't tolerate a totally polychrome environment. We just don't have the bandwidth.

Rail: No, we live in white boxes with some stuff on the wall.

Ritchie: And it's all embedded in a meaning system. This is the strangeness of the world we're entering. In my early work, I talked a lot about comic books and science fiction, and various detective stories, historical romances, other narrative genres that tell the story in a kind of understandable way. As in the nine-frame comic panel, you kind of know what's going to happen in a comic in a sequence. But you don't know what's going to happen on the internet. There's no sequentiality. What is that as an information environment? It's a manifold, a mathematical space. But what we haven't figured out at all is how we move in those dimensions, because we're still essentially living in linear time. We go onto the internet, and it's all human time at once. There's some research that the [Albert-László] Barabási lab did, that showed that by tracking people's cell phones, you can predict where people will be ninety-three percent of the time. On a daily basis. They're trackable, their routines over time. And it's the same in the information space. There are all these maps of the internet, and they show most people just kind of spiral around. They don't use what's really there.

Rail: There's very little continuity to what you see from screen to screen. And getting lost is not productive.

Ritchie: It's a condition of anxiety. But then you find, I guess, what I think is hopeful about that environment. There was a kind of comforting fantasy we all shared that the traditional library represented this kind of secure, safe space. But of course, libraries are biased and locked into their own narrative of presuming there's a hierarchy of knowledge, which can become exclusionary. They’re another form of spiral. To me, that's one potentially liberating thing about the space of machine intelligence. It's dehierarchizing itself. Twenty-five years ago, J.C. Herz wrote Joystick Nation about early video game environments, and Legacy Russell wrote another book recently called Glitch Feminism. Both are talking about a similar space, of not exactly, or not only, liberation, but one that does allow different degrees of freedom, that allows different perceptions to emerge in the battle-theater-commons of the internet. That jarring quality, that friction, is also obviously in my work, and it's also there in Bosch, Basquiat and Polke or Ensor. Skeletons, painting pictures, us.

Rail: What about the level of absorption of material that you experience through the internet? It seems that machine learning, such as it is, is like a sort of surface-based absorption. Is that one way that you're pushing back against this, by insisting on the primacy of your brush, your materials?

Ritchie: Well, that sounds very grand. "Insisting on the primacy of the brush."

Rail: We were talking about that this is a riposte to AI, and that idea of generating material without the human creative element.

Ritchie: I suppose, because this process is so fucked up, that it takes human generated material, and then turns it into computational space, and then prints that out. And then I must respond to that. The only way in the end I felt I could respond was just one hundred percent human. Because that's all I’ve got left.

Rail: Do you feel it freed you as a painter in a way that maybe you haven't painted in many years?

Ritchie: Roberta Smith, in her very first tiny, mention of my work, wrote, "conceptual artist, doing painting." I kind of took that to heart. She has that terrifying insight into your practice. You’re like, goddammit, I guess that's right. [Laughter] And for my generation in the UK , painting always had implied quotation marks around it, painting was kind of an assignment. Gilbert & George or Les Levine, or Mary Kelly, who was my teacher, were part of a continuum with the British Art & Language group who made big "paintings." There was this sense of a complicated discussion, a lot of suspicion of the free hand of the painter. Post-David Salle this was a thing that you thought about, it was a thinking medium. At the same time, I always loved, you know, Francis Bacon and Julian Schnabel and people like that, who just loved painting. I've always kind of existed in that duality. And then, smart artists whose work I know and love, like Cecily Brown, Julie Mehretu and Christina Quarles, just took it back to it being painting. But I was born into the flip side of that coin. So, with these new paintings I felt that if I can use machines to help me do the thinking about painting, because they're already all about the history of painting, the only thing that is left is to be a painter. Of course, I can't not paint the history of painting. It felt like an honorable way to finally surrender to painting. It won. Painting won.

Rail: Painting won. Again. It's been winning since 1400.

Ritchie: Since I think early 100,000 BCE, the first cave paintings.

Rail: Cecily Brown always says she had to leave England because she wanted to paint, that she didn't feel she could paint in London, that it wasn't being appreciated. In the late eighties, early nineties. So, she came to New York.

Ritchie: Yes, and even in England, Bacon and Freud, their mark was so constrained. You know, it's like a little spill, a little cross hatching. I think it's a literary culture, some preference for text over image. And for saying rather than showing.

Rail: There's a Victorian element to that.

Ritchie: Yes, the illustrative.

Rail: Here's my question about these paintings, about Generator and Discriminator (both 2022). Some of the other titles are Yew of the Wood and Battle of the Trees (2022). The titles feel arbitrary in a sense. Do they come to you later, or is it something that you’re going into it thinking about it? Or are these ideas of intelligibility and communication more important? They seem not latent, but present in these works. They’re trying to communicate. You have to kind of decode them, but it feels like you're leaving it to the viewer to do it to a degree.

Ritchie: Oh, one hundred percent.

Rail: One hundred percent! Entirely. Because you can say they relate to Renaissance art, and while Yew of the Wood resembles something, at the same time it doesn't, because it also looks like an aquarium. Now that you have talked about this central section [gesturing] as a kind of figure, I can see a hovering, maybe sort of avian element, but it feels incidental.

Ritchie: When I first did them, I thought it's the Four Heavenly beasts that support God in the chariot in the apocalypse. And then I was like, wait a second, which one is which? That doesn't make any sense. [Laughter]

Rail: But the one reference that did seem to be sort of running through a lot of this, both visually and thematically, is Max Ernst's Surrealism and Painting from 1927 in the Menil collection, in Houston. His image addresses the question of what would an amorphous shape, if it was a painter, make? And the answer is a sort of Kandinskyian abstract painting. [Laughter] And the secondary character, sort of in the center, I read as a muse-like figure. It's on top of a Man Ray-like Rayograph, a Miró-like box. It's such a strange painting, with this creature posing, with brush, having finished the painting.

Ritchie: Yes, almost posing as the artist.

Rail: And the figure posing there, atop the box?

Ritchie: Maybe that's the portrait. The artist seeing itself. That was how I always interpreted it. And the playing with the spray, and the kind of marks and the dots, these are the tracks of a hand. It's also very much about time. The figure is drawing this kind of atomic diagram, all these orbits. You’re wondering what the image is here? Is there actually another painting, inside the box? I saw Surrealism and Painting many times when I did a show at the Moody Center for the Arts a few years ago. It's a work of genius that reminded me of an observation people made when time was first articulated in Einstein's theory of relativity, which was that if you look at a human body through time it's a sort of tube of flesh, just moving. You start as like a tiny blob, and then you get sort of bigger. If you could see every frame of your life at once continuously in space it’d be this extended tube that swirled around and around itself, and then at the end sort of shrunk down back to nothing. I thought about Ernst a lot, both Surrealism and Painting and his Europe after the Rain (1940-42) when making Harbinger (2022). There's a similar kind of trick the surrealists did to include the blue sky, or a generic outdoors to make you feel it's alright to be inside their mind.

Rail: It's limitless.

Ritchie: There's something in that that relates to painting, smearing this quite gross, toxic substance across a surface. If you look at oil paintings close up, like those beautiful scans under electron microscopes, the thing that's really special is that the pigments don't dissolve, the crystals are maintained in there. They’re topologically distinct. Even when they’re laminated together. So, it's a multitude.

Rail: We see them as a seamless totality, but they don't exist like that.

Ritchie: And when I was making these sort-of-figure paintings, which I started slightly before the "Branches" ones in the larger rear gallery, I kept thinking about Cézanne, how he has that mark that just kind of goes on and on. It's the same mark all the time, yet slightly different. He's not van Gogh, he doesn't have a mark for everything. He's got ten marks.

Rail: He used passage. There's a real consistency.

Ritchie: And I felt this is like building a wall, a kind of a Jasper Johns-y thing as well, where you’re just like, mark, mark, mark. And then they accumulate into this semblance of an image. But it's also background as foreground.

Rail: How does this whole wall of drawings fit into all this?

Ritchie: I guess the drawings are versions of paintings that haven't been made. So I did them in very much the same method, via the program.

Rail: There are drawings that relate to each of the paintings?

Ritchie: They’re actually slightly different stops in the code. You can see they’re a little off. I wanted to show the absurdity of the program and its beauty. It literally makes millions of these, and you can just sort of pluck them out. They’re all titled Leaves.

Rail: Is there a print underlying these also?

Ritchie: Yes. You run the program and stop the program. I train it. Print the image. I get a kind of very vague faint hint of what the thing could be, and then I completely redraw it, in the most labor-intensive way any one person could do everything, using pencils, colored pencils, and inks. I was trained in a very classic, draw-from-the-cast art school. It's my go-to.

Rail: You were talking about how some of these come from decorative drawings at the Met, applied art drawings. And the loops and swirls and the movement out to the edge feels like that comes from a little bit more architectonic type drawings.

Ritchie: It started with the drawings at the Met. And they tend to be a sheet, with a thing in the middle. As I got more adventurous with the code, I started pushing it to do more and more ambitious things. The thing about this process, as anyone will tell you, is that eighty-five percent of coding is data management. I had to look at every drawing in the Met, 400,000 images, and I could never have done that without COVID. It was a gift of time. I had nothing else to do. Looking through the site, choosing them, and then putting them in little folders.

Rail: Can you talk a little bit about the project in a stairwell in the Bloomberg Center at Roosevelt Island for Cornell Tech? Everything that Rises Must Converge, 2017, resin, glass, and ink.

Ritchie: That was an invitation to do a work on the history of technology—because it is Cornell Tech, and I was able to refer to my vast collection of diagrams from a large work I made at the Getty, The Temptation of the Diagram (2017). I think the title is a Pierre Teilhard de Chardin quote. It's about the sense that although everything is changing, it's all accumulating and coming together rather than necessarily diverging and failing, a common theme in my work, a bizarre optimism in the face of ceaseless change. Although it can cause anxiety and even terror, we are all on the journey together. It's a sense of the journey as a positive one, wrenching and difficult though it may be. The title, Everything that Rises Must Converge, is kind of literal. You're going to climb this giant staircase that's the central axis of the building. People are up and down all day. They're going to converge, their thoughts will converge. The imagery is printed on a laminated layer between the glass. And the idea was that the students could join the history of technology. It was a puzzle I proposed to them. But you can solve it, a linear history of technology. You could do it with an erasable marker pen. Because they I noticed they were drawing all over the partitions.

Rail: I wonder if they're doing it.

Ritchie: They're not.

Rail: Because they won't let them because it's art.

Ritchie: Yes, central flaw in my plan. [Laughter]

Rail: You mentioned something tantalizing the other day. That all this somehow relates to drag. It emerges from an older work titled The Box Factory from Five of a Kind (2003)?

Ritchie: In that work, each character represents universal forces, like mass, space, and time and is dressed in a famous costume from the history of popular music. There's David Bowie's Yamamoto costume. There is a Bootsy Collins outfit and a Missy Elliott outfit, overtly engaging with this question of drag or masquerade. This is very much tied into my last decade of work with theater, musicians, and dancers, who for me represent the exploration of time through the embodiment of human presence, as in Infinite Movement with Shara Nova and my upcoming project with Hanna Benn and the Fisk Jubilee Singers for the Frist. There's a beautiful section in Judith Butler's Gender Trouble about what she calls "the styles of the flesh," the cultural production and swapping of gender identities, who's wearing which mask of gender, or power, and how the counter-identity is often both reflecting it and supporting it while being repressed by it. Being British, I always go back to Shakespeare, but you can find this in Molière or the harlequinade in Italy, or Grant Morrison's Invisibles, all stemming from the idea that the right mask can gift the wearer a new persona, which is from ancient Greek theater. The liberating feature of a large enough space of play, like the internet, is that everyone gets to try all the costumes in the costume box, all the masks. There are these constant reversals in the kingdoms, not only gender reversals but also power reversals. Which is also a very old idea, like Saturnalia, where the poorest person becomes the king for a day. The positive way we're exploring these spaces is also a kind of recognition of the absurdity of any one person occupying any space forever, which has been a flaw in how we thought about time. We put people in a box and they’re the king forever. You'll always be the king until of course you die. And then you're not the king, and everyone just pretends it's over. But there's an opportunity in the internet. That space is kind of anxious because many people don't want someone to not be the king. When you mentioned pushing back against artificial intelligence, it is definitely pushing back against the idea of a sort of tech billionaire, in their isolated compound, like that movie Ex Machina. You know, where it's sort of like, I'll build the perfect box. And I'll be here, being me, forever. Wherever you're stuck, to be stuck is the horror. Look at you and me, here we are in our exciting outfits. We're really enjoying the possibilities of sartorial freedom. Whatever we like, as long as it's dark blue, black, or gray. And pretty much a shape, but not too much of a shape. [Laughter]

Rail: Well, it has to be sensible, depending on if you're going from whatever A to whatever B?

Ritchie: Exactly, even if you are only a margin player, you could almost go to a funeral, or the gym, dressed like this and fit in. It's about a kind of vanishing, an abdication of the joy of an individual being a character. Using the GAN in the way I do here, is a way out of this convergence, or fitting, suggesting that like painting, information space can be a new kind of image or mask, built out of time. Being generated by the computer it seeks the idea of a generic behavior, but without thinking. So, it also doesn't have that kind of weight of x equals y, or so and so shouldn't be so and so, or so and so should be so and so, and as such it can be redirected towards a glorious confusion. That also has something to do with Creative Commons and mutuality of data access, but also it is a kind of de-weighting of the mask's symbolic power altogether. I think I've always felt that in digital space, there's the same opportunity that there is in any other creative space to do more than just fit in. That's what Butler's essay is about, that accepting the concept of the masquerade is ultimately freeing, because it makes you aware of the fragility and the beauty of everyone's roles, not just the pomp of power but the beauty of being a teacher, or a janitor. Rather than simulating the aesthetics of the corporate desktop, let's try to bring that back in, the kind of polymorphous perversity that underlies the creepiness of surrealism, but also the sublimity of Medieval and Renaissance art. I think the potential for joy and investigation in that kind of combinatory painting space is crazy. So, by the time these paintings arrived, I was ready to start thinking about paintings as pictures of time rather than occupying space. Strangely enough, this specific form of machine learning allowed me to admit that I was actually a painter, because it freed me from thinking, to be with the paintings in time. I kind of came out as a painter to myself. [Laughter] "Oh, I do like it!" But I think that a very important thing to articulate in the face of the kind of grimness of the conversation around AI is its potential freedoms. We are absolutely at an inflection point, and you can think of it as a liberatory space or as an imprisoning space. Those are both probably true. But if we just focus on the prison, then you never get free.

Jason Rosenfeld Ph.D., is Distinguished Chair and Professor of Art History at Marymount Manhattan College. He was co-curator of the exhibitions John Everett Millais (Tate Britain, Van Gogh Museum), Pre-Raphaelites: Victorian Avant-Garde (Tate Britain and the National Gallery of Art, Washington, D.C.), and River Crossings (Olana and Cedar Grove, Hudson and Catskill, New York). He is a Senior Writer and Editor-at-Large for the Brooklyn Rail.

James Cohan Jason Rosenfeld (Rail): Matthew Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Rail: Ritchie: Jason Rosenfeld