AI-Generated Art—From the 1980s
pple’s recent advertisement for the slimmest iPad was a crushing faux pas. Instruments of creativity—a trumpet, a set of oil paints, a drafting board, a stack of hard-bound books—crushed beneath the weight of a hydraulic compactor seemingly borrowed from a nearby scrapyard. (And you thought Apple’s “Mother Nature” advert was heavy-handed.) All tools of human artistic expression are now contained in an elegant piece of hardware may have been the metaphor Apple intended. It was not, however, the metaphor the ad relayed. Consumers everywhere saw something more sinister, a signal that the emerging AI-industrial complex will soon render your artistry as irrelevant as a 1982 Datsun. Tragedy is in the eye of the transgressed.
The age of AI anxiety and opportunism is upon us, labor’s cries of forthcoming Armageddon clashing with the cheerleading of techno-optimist investors and our collective gasping at the creative facility of tools like DALL-E. Long before the current hysteria, a quiet artist from Britain spent the better part of his accomplished career building an artificial intelligence software for art-making he named AARON. A timely exhibition at the Whitney Museum of Art placed Cohen’s ur-AI project on view, including a robot using his code to draw based on a series of prompts and a diverse selection of works produced by the computer program. The effect is at turns dreamy and disconcerting.
AARON is old-fashioned, which means he learned to draw without massive data sets extracted from the cloud to inform his composites. Instead, from the 1970s to 1990s, Cohen hand-coded instructions for objects, anatomy, botany, and relationships in a logic framework that itself constitutes a kind of functional art. The output is remarkable for such a process and wildly convincing that AARON is drawing from visual experience when, in fact, it is drawing only from Cohen’s semantics. Canvases are sometimes rudimentary, sometimes lofty, and always delightfully strange—but in an organic way that avoids the creepy uncanniness of today’s Midjourney. The quiet halls of the Whitney may levitate the works above their station, but the human figures, sketched independently by AARON and painted by Cohen, are deeply compelling. How can a machine make such decisions about length, setting, angle, and arrangement? How can it create coherence, composition, and even a sense of story from Cohen’s manual? “Legs,” Mark Zuckerberg once said of avatars in his absurd metaverse, “are hard.” AARON got them mostly right.
“If what AARON is making is not art,” Cohen once asked, “what is it exactly, and in what ways, other than its origin, does it differ from the ‘real thing?’ If it is not thinking, what exactly is it doing?” Cohen could not answer his question, and neither can we. Or at least not sufficiently. What we can answer is a simpler question: What was Cohen doing? Collaborating. With a tool more advanced than a brush, not yet as sophisticated as a large language model in a shared creativity that supercharges our capacity to imagine the artistic partnership between human and machine. Apple, take note.
