A.I. Pop Culture Is Already Here

We’re living in a world in which every style, every idea, and every possible remix can be generated as fast and frictionlessly as possible.
Illustration of snarling balloon dog that suddenly fractures.
Illustration by Nicholas Konrad / The New Yorker

Last month, a YouTube user named demonflyingfox uploaded a video titled “Harry Potter by Balenciaga.” It showed characters from the Harry Potter films—Hagrid, Ron, Hermione, Snape, McGonagall, Dobby—as gaunt models with aggressive cheekbones (slightly yassified), dressed in gothic capes and leather jackets. Set against a catwalk-worthy electronica beat, the actors blink, nod, and speak lines from the books which have been remixed with fashion references. “You are Balenciaga, Harry,” Hagrid says, instead of breaking the news that Harry is a wizard. The video is strange and hilariously sinister. In three weeks, it has received almost five million views; a sequel, released less than a week ago, has netted more than a million and a half. Pop-culture mashups of one famous thing with another are an archetype of Internet meme-making. What’s unusual about “Harry Potter by Balenciaga” is that it was generated with artificial-intelligence tools. As the video’s creator, the Berlin-based photographer Alexander Niklass, who made the demonflyingfox channel, told me, the video demonstrates a newfound ability of A.I. to “create filmlike moments.”

A.I. tools were involved in each step of Niklass’s process, and in each element of the video. He created the basic static images with Midjourney, evoking the Harry Potter actors and outfits through text prompts such as “male model, grotesque, balenciaga commercial.” Then he used ElevenLabs—a “voice-cloning” tool—to create models of the actors’ voices based on previously recorded audio. Finally, he fed the images into a service called D-ID, which is used to make “avatar videos”—subtly animated portraits, not so far off from those that appear in the newspapers of the Potter world. D-ID added the signature lip synchs and head nods, which Niklass explained were a reference to fashion models tilting their chins for the cameras. The combination of child-friendly film and adult luxury fashion neither held any particular symbolism nor expressed an artistic intent. It’s “entertainment,” Niklass said. Yet the video’s most compelling aspect might be its vacuity, a meaningless collision of cultural symbols. The nonsense is the point.

A.I. tools may have been able to replicate actors’ faces and generate fashionable outfits, but only Niklass could have come up with the concept, which required keen observation of both high fashion and the wizarding world—and also a very specific, extremely online sense of humor. With tools like Midjourney publicly available to anyone online, “everybody can create something visually appealing now,” he said. “But A.I. can’t generate taste yet,” he continued. By “taste,” he meant “a good aesthetic judgment”—background knowledge of what you’re generating and a sense for what looks good, without falling too far into the uncanny valley. To put it another way, execution may have been democratized by generative A.I., but ideas have not. The human is still the originator, editor, and curator of A.I.’s effects. Proof of Niklass’s taste can be found in the many copycats of his videos now on YouTube: Anyone can access the same technology and attempt a replica following the formula he set. (A video tutorial made by an A.I.-education channel called PromptJungle shows the exact process.) There are “Matrix by Gucci,” “Star Wars by Balenciaga,” and “The Office by Balenciaga” videos, but none of them are as appealingly odd as the original.

While no one would mistake “Harry Potter by Balenciaga” for real footage—except, perhaps, as a real fashion-advertising campaign—another A.I.-generated image recently made news headlines because so many thought it was real. It was, or appeared to be, a photo of Pope Francis walking on the street wearing a baroque version of a puffer jacket, silken white, with a high collar and a hood. A cross necklace swung from his neck and he carried a to-go coffee. “Swagged out” would be an appropriate description. The detail and texture of the image made it appear utterly realistic. When I first encountered it, while quickly scrolling by, I thought it was real and almost unremarkable—of course the Pope has fancy clothes. The model and actor Chrissy Teigen tweeted that she “didnt give it a second thought.” But it was created using Midjourney by a Chicago man who identified himself as Pablo Xavier when he was, as he told the Chicago Tribune, high on mushrooms. He used prompts with phrases such as “Catholic Pope Francis. Balenciaga puffy coat. Streets of Paris.” (Perhaps the fashion brand is primed for A.I., or it comes easily to mind as a signifier of avant-garde luxury.) “I just thought it was funny to see the Pope in a funny jacket,” Xavier told BuzzFeed News.

Created without much intention besides an imaginative whim, the images were so potent because they were made to look so photographic, a capability now in the hands of anyone online, not just highly skilled human photo retouchers. A.I. automates creative impulses, negating the labor involved in producing an image or a video. (A real version of the “Harry Potter by Balenciaga” videos would likely have cost millions of dollars in talent alone.) Mulling over this discrepancy between input and output, I had a dialogue with ChatGPT. It was more Socratic than an actual discussion, a bit like talking to the mirror. But it helped me elucidate my own reactions. I asked how A.I.-generated imagery was changing our perceptions. It responded that there has been a “blurring of the lines between real and artificial.” Then I asked, Isn’t it true that even an A.I.-generated, “artificial” image is also a “real” thing, because some human caused it to be made, as a kind of cultural wish fulfillment? The robot responded that the realism of A.I. images “is often designed to be illusory.” This idea of “illusory realism” struck me as apt. The A.I. content has the appearance of realism, without actual reality—reality solely as a style.

In a 2022 interview, David Holz, the founder of Midjourney, used the phrase “aesthetic accelerationism” to describe the profusion of generated imagery enabled by public A.I. tools. It evokes a world in which every style, every idea, and every possible remix is generated as fast and frictionlessly as possible, and the successful ones stick and get attention, like “Harry Potter by Balenciaga” and the swagged-out Pope. It may be less because they are artistically great than because they solved some formula of attention. Perhaps the successful creation is unimaginably bizarre, the seamless merging of two unrelated things. Or it’s driven by the fascination of the perfect replica, something that we know isn’t real but which is easy to see or briefly perceive as such, like a trompe-l’oeil painting.

For the past few days, I’ve been looping a new hip-hop track over and over. It’s called “Savages,” by the French outfit AllttA. The song is sweetly nostalgic, with synthesized strings and a snare backbeat; it features what sounds like Jay-Z trading verses with AllttA’s Mr. J. Medeiros in a throwback style. But, of course, it’s not Jay-Z; it’s an A.I. model of his voice, used, presumably, without the artist’s permission. It’s another example of illusory realism. The human-written song is good on its own, and would be perfectly fine without the fake Jay-Z, but the familiar voice adds something ineffably compelling to the track, making it sound like an unreleased B-side from the nineteen-nineties. It has more than two hundred thousand plays on YouTube. “The thought of enjoying this and it’s AI is beyond me,” one user wrote in the comments. I feel the same kind of existential confusion. It sticks in my brain like an unsolved puzzle. I don’t care that it’s not actually Jay-Z, in large part because the A.I. quality is good enough that I, a non-expert, can barely tell the difference. But it seems that a Rubicon has been crossed: It doesn’t matter that these artifacts are generated by A.I.; we can just enjoy them for what they are. It happened faster than I thought possible, but now that A.I.-generated pop culture has entered the mainstream, it seems unlikely that we’ll ever get rid of it. ♦