Over the winter holiday, I noticed William Gibson‘s Mastodon account went quiet after a few posts. No explanation, but like many people who seemed ready to leave Twitter after Elon Musk purchased it, he reverted to his old account where I found this exchange.
Twitter has devolved entirely into virtue-signaling, regardless of one’s political position so, initially I thought that Gibson was simply agreeing, but this also seemed odd (“weird as hell?”) coming from him since his Neuromancer trilogy expressly addressed the creation of art by Artificial Intelligence. Take Count Zero, in which the Boxmaker, an AI in a forgotten high-orbit space station, assembles works in the style of Joseph Cornell so convincing that they are taken as real before being revealed as forgeries. Just possibly, they might look something like this.
Of course, in its present form, an AI image generator* is a tool, not so much an Artificial Intelligence, but rather a desiring machine—an algorithm that combines images in ways it predicts will satisfy us. AI image generators have no inherent intelligence or understanding of how these things go together, they are nothing more than programs that desire connection in ways that have been conditioned over time, “cognition” at the level of an insect seeking a red flower or perhaps a virus “seeking” a host. This is obviously, very different from Joseph Cornell, an artist of deep talent, capable of creating works that fostered emotional connection.
But what about the Boxmaker in Count Zero? On the one hand, the Boxmaker is capable of making works that are supposed to be enthralling. Take this passage from the novel in which disgraced art dealer Marly Krushkova (disgraced because she tried to sell a work by the Boxmaker as one of Cornell’s) encounters a hologram of one of the Boxmaker’s works.
… she took the package to the window and turned it over in her hands. It was wrapped in a single sheet of handmade paper, dark gray, folded and tucked in that mysterious Japanese way that required neither glue nor string, but she knew that once she’d opened it, she’d never get it folded again. The name and address of the Galerie were embossed in one comer, and her name and the name of her hotel were handwritten across the center in perfect italic script. She unfolded the paper and found herself holding a new Braun holoprojector and a flat envelope of clear plastic. The envelope contained seven numbered tabs of holofiche. Beyond the miniature iron balcony, the sun was going down, painting the Old Town gold. She heard car horns and the cries of children. She closed the window and crossed to a writing desk. The Braun was a smooth black rectangle powered by solar cells. She checked the charge, then took the first holo fiche from the envelope and slotted it.
The box she’d seen in Virek’s simulation of the Güell Park blossomed above the Braun, glowing with the crystal resolution of the finest museum-grade holograms. Bone and circuit-gold, dead lace, and a dull white marble rolled from clay. Marly shook her head. How could anyone have arranged these bits, this garbage, in such a way that it caught at the heart, snagged in the soul like a fishhook? But then she nodded. It could be done, she knew; it had been done many years ago by a man named Cornell, who’d also made boxes.
For Krushkova, the Boxmaker’s works are evidence not of a trained machine but of the creation of aura, emotional connection, and nostalgia from simple objects—something far ahead of any mere Turing test, a spark of the divine at work. Strangely, it’s that the Cornell boxes are convincing enough to be forgeries—e.g. not original art, but rather pastiches of no originality, the work of the academic, the dilettante, the poseur, the forger—that is, for Gibson/Krushkova, somehow evidence of greatness. This doesn’t ever seem to be explained or resolved in the novel, but in an interview three years after the book’s publication, Gibson explains why he has the Boxmaker in Count Zero copy Cornell:
WG: If I was doing a thesis on my work, I would try to figure out what the fuck that Joseph Cornell stuff means in the middle of Count Zero. That’s the key to the whole fucking thing, how the books are put together and everything. But people won’t see it. I think it actually needs someone with a pretty serious art background to understand it. You know, Robert Longo understood that immediately. I was in New York—I’ve got a lot of fans who are fairly heavy New York artists, sort of “fine art guys,” and they got it right away. They read those books around that core. I was actually trying to tell people what I was doing while I was trying to discover it myself.source: Darren Wershler-Henry, “Queen Victoria’s Personal Spook, Psychic Legbreakers, Snakes, and Catfood: An Interiew with William Gibson,” Virus 23, Issue 0 (Fall 1989): 28-36.
DWH: It goes back to postmodernism, to pieces again, and to making new wholes from fragments, doesn’t it?
WG: Yeah. It’s sort of like there’s nothing there in the beginning, and you’re going to make something, and you don’t have anything in you to make it out of, particularly, so you start just grabbing little hunks of kipple, and fitting them together, and … (laughing) I don’t know, it seemed profound at the time, but this morning it’s like I can’t even remember how it works.
In another interview, Gibson states that he decided to write books in homage to artists he “particularly loved or admired.” (“William Gibson Interviewed by David Wallace-Wells, The Art of Fiction,” Paris Review, No. 211). Perhaps Gibson choose Cornell because the artist’s boxes recapitulate cabinets of curiosities (Wunderkammer), assemblages of disparate and unusual objects from around the world, intended to provoke a feeling of wonder and enchantment. The famous Musei Wormiana, for example, has turtles, shells, armadillos, crocodiles, deer heads, a narwhal tusk, and other strange things on the walls, while numerous terrifying fish, a bear, and a kayak hang from the ceiling.
This seemingly random assemblage, however, is actually a network of juxtapositions and comparisons intrinsic to the methodology of “natural philosophy” or “philosophy of nature,” a precursor to the scientific method that sought to explain philosophical truths about the world—the macrocosm—by looking at objects—the microcosm. Cornell himself studied natural philosophy from antiquarian books and, like its practitioners, sought kinship between seemingly diverse elements and to produce wonder about the order of the macrocosm from the microcosm.
But cabinets of curiosities also played a particular world-historical role, indicating their assembler’s ability to obtain items from exotic lands in a newly globalizing market. The wonder the Wunderkammer provokes is also wonder at the reach of colonial trade and the ruling classes’ mastery of the world. The things Cornell found, in contrast, were generally obsolete, things could be considered junk, detritus of a recently-bygone civilization, not precious objects but rather cast-offs among the bric-a-brac dealers and junk shops of mid-twentieth century New York City.
Unlike the works I have produced for this project using Midjourney, an AI image generator, Gibson’s Boxmaker makes art out of physical materials. Nevertheless, contra Cornell, the Boxmaker does not gather materials from junk dealers, rather it returns to the original use of the Wunderkammer to display surplus wealth: this is detritus, but from the high-orbit attic of a fabulously wealthy family—leftover mementos from the Tessier-Ashpool clan—”a yellowing kid glove, the faceted crystal stopper from some vial of vanished perfume, an armless doll with a face of French porcelain, a fat, gold-fitted black fountain pen, rectangular segments of perf board, the crumpled red and green snake of a silk cravat…” Such items would have been fabulously expensive by the 2040s, the decade Count Zero is set in. Wonder, in this world, would be triggered by the thought that such things still exist and that someone would have the wealth to use them to create art, much as one might wonder at finding an early illustrated book used by nineteenth-century children to practice coloring in.
Back to the Wershler-Henry interview again. Gibson talks about the moment when Marly sees the Boxmaker: “Marly follows the map in that book. She’s the only one who can receive the true map and she goes to the heart of it. She gets an audience with God, essentially, and she does it through her own intellectual capacity and her ability to understand the art.”
The Boxmaker, however, is rather like Midjourney, not a general intelligence but a device limited to one task. The “true artist” is not the Boxmaker, it is the Boxmaker’s creator Lady Jane, an eccentric, “stone crazy” last heir of the Tessier-Ashpools. As an off-screen character in Count Zero, her motives are obscure. The forgeries make their way to the market through an underground network full of profit-taking and she sees nothing from them, so why make them except as a wealthy exercise in dilettantism? Cornell thought that viewing his boxes could change lives, but none of this is important in the autistic world of Lady Jane and the Boxmaker. How, then, to reconcile this with Krushkova’s reception of the boxes? Because, frankly, it seems weird as hell.
Today, we live in an entirely different cultural and computational regime from the mid-1980s (let alone the mid-20th century). Generations have grown up replicating Cornell’s boxes at educational events at museums, in art classes, and for DIY interior decoration. Cornell’s boxes are over-exposed and oversaturated to the point that for many people they have been emptied of wonder themselves. But maybe there is something to Gibson/Krushkova’s awe at the forgeries? Looking for Cornell’s message in his work would be a mistake and there is no meaning behind the (unseen) works of the Boxmaker or behind the work I generated with Midjourney (although crucially, I did intervene as an editor, selecting only some of the images Midjourney produced). And yet, if there is anything that provokes wonder about the works on this page, perhaps it is that we are gazing upon the collective unconscious of humanity as embodied in the datasets. Could AI image generators be a way—not of replacing the artist—but of creating both a new kind of Wunderkammer and a new Surrealism? Perhaps what I am really after is trying to make Cornell look new again, to defamiliarize his world just as he sought to defamiliarize the consumer culture of midcentury America. And making everything “weird as hell” (e.g. ostranenie) again is exactly what art is for.
* My friend Lev Manovich points out that referring to these image generators as art generators is a mistake. This is correct. Midjourney, Dall-E, and Stable Diffusion are not art generators, they are image generators, just as a camera does not make art. They can be used to make art, but that depends on who is manipulating them.