The CFDA wants to help designers navigate generative AI

A first-of-its-kind partnership between the CFDA and AI platform Raive both protects designers from AI and guides them to using it.
Image may contain Alice  Olivia Clothing Long Sleeve Sleeve Dress Formal Wear Adult Person Fashion and Gown
Photo: Alice + Olivia

Sign up to receive the Vogue Business newsletter for the latest news and insights on AI in fashion, plus exclusive membership discounts. To become a Vogue Business Member and receive the Technology Edit newsletter, click here.

The Council of Fashion Designers of America (CFDA) has partnered with artificial intelligence startup Raive to help its more than 400 members make sense of generative AI as it circles in on the fashion industry.

Members will have access to education and pilot programmes that will test uses of generative AI, which can create images, videos and text. The goal is to both protect designers from having their work used without attribution or compensation, and teach them how to use AI to their advantage.

“Brands and designers are always looking for new tools and processes that can expand creative boundaries or streamline their design processes, and AI lends itself nicely to those interests,” says CFDA CEO Steven Kolb. At the same time, he adds, there is still some fear about AI, especially regarding intellectual property and the future of jobs and creativity, so this programme is in part an effort to answer questions and alleviate fears.

Kolb says that compared to a number of recently popular tech experiments, such as NFTs and the metaverse, the practical, present-day applications of AI have “a bit more clarity”. He sees it as a long-term, transformative shift. “E-commerce didn’t go away, social media didn’t go away, and this is one that will resonate for a while.” And there is clear interest from CFDA members: an early virtual professional development programme was one of the best-attended programmes in recent history, Kolb says.

Raive's software enables designers and marketers to create and edit scenes with text prompts, then edit or add in specific elements, including product imagery.

Photo: Raive

Three CFDA members including Alice + Olivia have signed on to work more closely with Raive to develop internal AI strategies; learnings from these pilots will ultimately help inform a blueprint that can be shared more widely with additional members.

Alice + Olivia’s VP of IT Ira Miller compares the advent of generative AI to Amazon’s imminent rise. “You know it’s coming, you know it’s disrupting the industry and your choices are to ignore it and put your head in the sand or find ways to embrace it and empower your business to take advantage of it,” Miller says, adding that Alice + Olivia’s leadership team is embracing AI as a “power tool” for designers. “I like to think of generative AI as a ‘bionic arm’ that speeds up productivity for our various teams, allowing them to focus on their core competencies where they provide the most value.”

Raive, founded in 2022, is a foundational AI company, meaning that it has created a general model trained on broad data, which can create new works based on image and text-based prompts. A competitor of companies including Midjourney and Runway, a key differentiator is that it is designed to attribute when it uses specific intellectual property and to then provide royalties to official licensed IP holders. Participating brands can elect to ingest their IP into Raive’s system, meaning that in the future, if someone creates something on Raive that uses a brand’s IP, the brand could be compensated.

For creative industries like fashion, this provides a potentially welcome solution to the concern that AI systems will use protected works to create new works, without credit or compensation. (For example, AI-generated images of people wearing Balenciaga and Nike recently made waves, with some confused if the brands participated — which they didn’t.)

Raive also enables brands to create an in-house image model that isolates its intellectual property from the base model, meaning that brands can design artwork that specifically only uses the brand’s own data. They can also create a scene with a text prompt, then add in a product image that they have submitted.

Raive is a general-purpose generative AI tool designed to compensate IP holders.

Photo: Raive

McKinsey & Company estimates that generative AI could add anywhere from $150 billion to $275 billion to the apparel, fashion, and luxury sectors’ operating profits within the next three to five years — primarily aiding in augmentation and acceleration, rather than automation, its analysts report. Designers and brands have experimented with a range of uses. Marc Jacobs used it to write show notes, while Hillary Teymour, of Collina Strada, and Olivier Rousteing, of Balmain, have tested remixing past designs into future work, with varying degrees of success. Teymour ultimately produced hand-edited versions of the generated designs, while Rousteing said that “my designers in my team could have done better. It was really good, but not as good as what we could have done on our own,” while speaking at March’s SXSW conference.

Norma Kamali is also training an in-house model to generate future designs, and retailer Revolve sponsored a design competition to produce future styles. More practically speaking, those including Estée Lauder Companies and Adore Me are using it to inform marketing and website copy, while Kering and Rent the Runway are testing shopping assistants.

There remain concerns over the threat to jobs; Levi’s drew controversy over its AI-generated ecommerce models, and fashion models are arguing for more protections over re-uses of their imagery. More broadly, media companies are already pursuing legal action against general models that train on published works without crediting, or compensating, the sources.

The education provided by Raive will help brands test the economic and creative opportunities provided by generative AI, and will focus on brand protection within the fashion industry, says Raive co-founder Sally Shin. Some CFDA member brands will participate in workshops and one-on-one advising, in addition to pilot projects. The brands who are part of the partnership will have access to their own custom image model, called Raive Enterprise.

Alice + Olivia plans to generate unique generative AI models for multiple specific product categories by feeding in historic designs; the design team can then use this to develop new products, which hopefully will accelerate the time to market, Miller says.

Miller hopes that it will help Alice + Olivia’s design team more quickly iterate through various inspiration paths, with the designers taking over on final designs and tech specifics once they decide what to produce. “Generative AI will allow our design team to focus on what they love doing — translating sketches and inspiration images into actual styles to produce — and help compress the design cycle,” Miller says, adding that upskilling will provide critical professional development for the team.

Already, after exploring tools including Adobe’s Creative Cloud Suite, the creative team has found that this area of generative AI is more concrete in terms of productivity enhancements, as well as protection of IP and other legal concerns, Miller says. For the corporate site, who are testing tools including Microsoft CoPilot, ChatGTP and Claude, the uses aren’t totally ready for “primetime” in part due to legal and data privacy concerns (it does help with functions such as emails and research).

There are future opportunities to, for example, enable customers to “remix” pieces within the brand’s own parameters, or dress their selfies in items that have been designed specifically only on the brand’s own works. While the promise of royalties holds considerable appeal, the first step is education, Shin says.

“Instead of ‘genie in a bottle’, we will have ‘Alice + Olivia in a bottle’,” Miller says. “The whole team is excited to see where it takes us.”

Comments, questions or feedback? Email us at feedback@voguebusiness.com.