Where is the biggest gap between the AI story fashion tells and what's real? Adoption is concentrated at the two ends of the product funnel: inspiration and creative design at one end, marketing and content at the other. But fashion doesn't only sell images and ideas, it sells physical products that have to be developed, sourced, priced and produced, and there's far less AI in that middle. "The story fashion tells itself is: we're really good at coming up with ideas and presenting those ideas. The gap is what it takes to actually make garments."
Ben calls online shopping "buying a fiction." What does AI change? By definition, shopping online means buying images, not touching products. People always accepted some fiction: better makeup, better skin, retouched photos. What changed is the scale of it. The answer, in his view, is accuracy: generative workflows "have got much, much better about representing real products," and the brands that win are the ones whose imagery matches what arrives in the box.
AI is cutting production costs. Where does the money go? Not into cheaper products: "retail prices, once they go up, don't come down." The honest version of the story is different. Traditionally, scaling revenue meant scaling content costs in lockstep. AI separates the two, so a retailer can grow without commensurately inflating overheads. That, he argues, is the story brands should actually be telling.
Will AI labeling laws fix the trust problem? His test case: if Netflix had to disclose every production that used AI anywhere, the label would be on every show within a year, "and if it's on everything, it's not handy." New York's synthetic performer law is purposefully broad; the EU AI Act is fixated on deepfakes and, as he reads it, leaves loopholes. The desire to label comes from the right place. Practical mid to long term? He doesn't think so.
What's the most underrated use of AI in fashion? A conversational layer over the enterprise stack. Connect PLM, ERP and the rest, let anyone in the organization query them in natural language without a full license seat. "The technology's there to do it... there are companies working on it, but I don't hear people talk about it enough."
Ben expected fashion designers to rebel against AI. They didn't. Why? He fully expected fashion design to push back the way music, film and publishing did, and it never happened. Designers work high-stakes and high-throughput on compressed calendars, and generative AI supports people who are overstretched. From the first AI Fashion Week in 2023, which he thought devalued the craft, to today, creative design has become one of the two biggest AI use cases in fashion. In his words: "Did not expect that."
Models are turning up in AI imagery they never consented to. What fixes it? Contracts. In music and publishing it's long established that you can't take paid work and riff on it without compensating the person, and modeling contracts will catch up with what generative AI makes possible. Then he turns the question around and asks how Botika handles it, and Jess answers on air: every Botika model is built from scratch, every feature generated, no scraped faces, no reference-image inputs, and the brand gets full copyright. His verdict: "Sidestepping it the way that you have is probably a neat choice."
