Fashion Images’s AI Reckoning

Last 12 months, a J.Crew marketing campaign selling the corporate’s collaboration with Vans roiled certain corners of the web involved with the sanctity of menswear. The photos appeared at dwelling for a model constructed on the signifiers of blue-blooded Americana: a barrel-chested WASP lolling on his daysailer, his flaxen hair swept again in good dishevelment, his jawline chopping good Harvard crew-team symmetries. Except there have been some peculiarities. The contours of that jawline had dramatically realigned themselves within the time it took for him to dock. Other conspicuous glitches: the stripes of a rugby polo fuzzed into static, a foot implausibly torqued. The initially uncredited marketing campaign was judged to be AI-generated. J.Crew ultimately admitted it was the work of Sam Finn, a London-based AI picture maker who, with presumably little irony, goes by “AI.S.A.M.”

To devotees of the mall model’s “vintage” Eighties aesthetic (Ivy League, Nantucket, Caucasian) referenced by these photos, the selection felt like a betrayal. Here was a model cannibalizing its personal historical past, sloppily regurgitating it and feeding it again to shoppers like some malformed lunch meat. Worse, it was a machine making a mockery of their nostalgia. Their self-styled style was revealed to be so flat as to be replicable by algorithm.

“I think the thing that mostly annoys people about AI is it’s sometimes really good and sometimes really bad,” Charlie Engman, a photographer whose engagement with AI straddles business and artwork practices, advised me. “It fails a lot, and I think the failures are actually very instructive and tell you what we care about. And then when it succeeds people get upset because then they feel manipulated.”

The debacle was much like one a month earlier involving a Guess ad within the August version of Vogue, a two-page unfold that includes a lissome mannequin wearing two seems to be from the model’s summer season assortment. Here, too, there have been inconsistencies: The mannequin’s face subtly morphed throughout the pictures, her pores and skin giving off the plasticized matte end native to residents of the uncanny valley. This time, although, Guess wasn’t making an attempt to idiot anybody. In small print on the prime of the web page appeared the phrases “Produced by Seraphinne Vallora on AI.”

In each these instances, the pictures had been sensible, that’s, they spoke within the language of images—naturalistic lighting, figures that credibly learn as human. Objectors (there have been a lot) famous the bitter irony of billion-dollar corporations undercutting expert labor to supply one thing that appeared like what they usually do anyway. In reality, that was much less ironic than the complete level.

The concept that promoting is unscrupulous is by now nicely understood. Fashion promoting specifically has a well-recorded historical past of emotional and technical deception. More than a decade in the past, Photoshop prompted an epistemological disaster—all of the sudden the waistlines of fashions in promoting and shopper magazines shrank to unnatural levels, and cheek fats dissolved with the swipe of a cursor, leaving a lot physique dysmorphia within the wake of the therapeutic brush instrument. Even celebrities—the already lovely—weren’t proof against the retoucher’s gaze. The boundaries of visible actuality had been compromised. The public discovered that images might now lie, and possibly had been mendacity.

The vogue picture’s elastic relationship with reality stretches additional again than the appearance of Adobe software program. Artifice was a part of the deal from the start. Edward Steichen, the godfather of economic vogue images, launched tips of lighting, focus, and lengthy publicity to control the feel of cloth or flatter a face. As expertise has turn into extra subtle in pursuit of the identical objective, AI’s whole artificiality is, in some ways, extra sincere. Andrea Petrescu, the twenty-five-year-old cofounder of Seraphinne Vallora—the advertising company behind the Guess advert—articulated, maybe inadvertently, a philosophical loophole. “We don’t create unattainable looks,” she told the BBC. “Ultimately, all adverts are created to look perfect and usually have supermodels in, so what we are doing is no different.” The vogue {industry}, lengthy maligned for selling unattainable requirements of magnificence, would appear to succeed in its logical endpoint in AI: a regular of magnificence that can not be accused of being unattainable, as a result of the gorgeous individuals don’t exist.

The concept that promoting is unscrupulous is by now nicely understood. Fashion promoting specifically has a well-recorded historical past of emotional and technical deception.

From a world-historical view, AI is just the most recent entry in industrialized automation. The rising accessibility of AI instruments means what as soon as took dozens of specialised roles can now be achieved (or approximated) by one particular person clicking round in Midjourney. Some of probably the most conspicuous makes use of of AI in promoting have been commissioned by multimillion-dollar corporations, which by their nature should adhere to capital’s ruthless logic. Finn, for example, is credited with AI imagery for mass-market attire manufacturers like Ugg and Skims that’s, like his J.Crew work, comparatively tame in comparison with the nightmarish stuff he’s cooked up for Alexander McQueen (gargantuan beetles) and Gucci (a psilocybin-soaked AI sequence shoehorned into final 12 months’s The Tiger, a weird fever-dream-as-brand-showcase directed by Spike Jonze and Halina Reijn and starring Demi Moore). For manufacturers that want to telegraph an edgy persona, AI reads like shorthand, if not obligatory.

Laura Dawes, a director on the London workplace of Webber Represents, a inventive company that manages photographers in addition to stylists, set designers, and artwork administrators, advised me that navigating demand for AI is an evolution of what she’s at all times achieved as an agent, which is to handle expectations. “The thing about artists is that everyone is so unique,” she mentioned. “I’ve had some really embrace it, and that’s usually in a controlled studio setting, something that’s more, say, constructed. But then in an uncontrolled environment, with natural light, that can be quite hazardous.” Dawes mentioned she has seen purchasers use AI to construct mockups for campaigns, which invitations an insupportable degree of threat: “You’re making an artificial image and using that as a guide and selling that to stakeholders on set when it’s not physically possible.”

Commercial adopters of AI are seduced much less by the expertise’s supposed mystique than by its labor efficiencies. It’s a brief leap from quick vogue’s already algorithmic enterprise to AI’s promise of value optimization. A recent report from the International Advertising Bureau tasks that generative-AI inventive will attain 40 % of all adverts starting this 12 months, confirming what anybody who watches YouTube movies or repeatedly rides the subway might already predict. It’s straightforward to see why AI would attraction to J.Crew, an organization that filed for chapter in 2020 and appears to be working by way of a protracted identification disaster. “In a way, you can’t fault companies like Guess or J.Crew for doing this, because it’s just following profit models that have always existed,” Engman mentioned. “Having worked in a lot of these jobs, you know, the creativity is very marginal. From a conceptual standpoint, it’s like, yeah, sure, make the J.Crew catalog.”

“Like it or not, this technology is here,” mentioned Kalpesh Lathigra, who teaches within the MA commercial-photography course at London College of Communication. Lathigra, an artist and documentary photographer who additionally works commercially, believes that the {industry} will see what AI can do, however extra importantly, what it can’t. “Personally, I’m not interested in AI. I would much rather get out into the world. A machine can give you millions of possibilities, but it can’t give you that elusive, intangible thing that draws us in and holds us, which has remained the same since the dawn of photography.”

Photographers and fashions, the image-industry jobs with the biggest cults of persona, could also be at much less threat than much less seen technical roles. “The people that I’m most worried about are the set designers,” Engman advised me. “That’s the job that I feel is already disappearing.” Those roles are being supplanted by the elevated presence of retouchers, a few of whom have cannily rebranded themselves into what Engman refers to as AI consultants, “which is a very interesting thing, because historically, retouchers were not seen as a particularly creative part of the process.”

There’s common resignation to the concept that AI is hastening entropy in a sure a part of the medium. “The race to the bottom I feel will only apply to the basics of image-making—what was once referred to as ‘pack photography,’ home catalogs, basically, but today is product imagery for e-commerce,” Lathigra advised me.

Dawes agrees. “When friends of mine were entering the industry, there were a lot of entry-level gigs mainly based around e-commerce, where people could do those kinds of jobs and have the ability to shoot their personal work on the side,” she mentioned. That was solely a decade in the past. “Now there are brands producing all of their e-com in AI. That for me is quite scary, because that’s such a large section of the industry.” Dawes mentioned she now routinely offers with purchasers whose total catalog of product images is AI-generated.

Engman, who has shot for Prada and Gucci, was an early AI adopter, beginning to mess around with AI fashions in 2022. He has an openness towards its prospects that different photographers won’t. It’s an openness that has made him sought-after within the vogue world as an AI oracle; he appears to speak about AI as a lot as he makes use of it. A few years in the past, he estimated that 80 % of his work engaged AI not directly, like a marketing campaign for Acne that featured statues of vaguely humanoid figures who had appeared to ossify into seashells whereas purchasing. He places his present AI output vis-à-vis commissions a lot decrease. Did he get bored, or is the flawlessly fashioned floor of AI’s bubble already deflating?

“I felt like I was participating in the bubble a few years ago, and now I think we’re in the awkward growing pains of moving out of the bubble,” he advised me. Engman recalled a current shoot with Coach, which had commissioned him primarily based on his AI work however then realized that wasn’t really what it was interested by. “What they wanted to do was much easier and more effective to do in CGI,” he mentioned.

The style of AI has already coated our mouths sufficient that it might not matter if it’s really current in any respect. Vogue’s December 2025 cowl options a picture of the actor Timothée Chalamet wearing Celine denims, cream topcoat, and untied motorbike boots, inexplicably posed on prime of a swirling nebula. It calls to thoughts the kitschy roll-down backdrops of college image day. The portrait, shot by the perennial Vogue contributor Annie Leibovitz, isn’t AI, however with its disproportionate scaling and goofy premise, has the identical acrid aftertaste. In some ways, that’s worse. If we’ve reached the purpose the place people are aping machines aping people, particularly unconsciously, we’re additional down the valley than we notice.