DIVERSITY IN PIXELS: CAN AI ERASE OR ENHANCE REPRESENTATION IN FASHION?
By: Erica Commisso
In every aspect of our lives, we see AI taking over. From chatGPT to X’s Grok, from AI-generated images to self-driving cars and musical mashups, the technology is taking over. But just how far will it go in fashion? Can AI replace models and will it enhance diversity?
In March 2025, Levi's unveiled a partnership with Amsterdam startup Lalaland.ai, which aimed to diversify and supplement the brand’s website with AI models. According to the Levi’s website they claim that it would use the models to increase "the number and diversity of our models for our products in a sustainable way."
Fashion retail giant H&M also announced earlier this year that it would use “digital twins” of its human models, adding that it would work with agencies and the models themselves to carry out the use of these digital twins in a responsible way. In some emerging applications, models can retain ownership of their digital replicas—enabling them to license their likeness to multiple brands and earn income each time their virtual twin is used, much like they would in a traditional campaign.
Source: Jon Tyson / @jontyson
Of course, that decision has come with some backlash. Using AI to replace human models can trigger significant backlash from several directions, especially when it comes to issues of ethics, equity and authenticity. It also raises concern over the future employment of makeup artists, hairstylists, stylists, photographers and all the other people typically hired to work on photo shoots.
Digital models like Shudu, the first virtual supermodel, and Miquela Sousa, a CGI influencer with over two million followers, are reshaping how fashion brands think about representation and cost. These figures don’t age, don’t make mistakes, and are infinitely customizable. For brands, this kind of control is alluring—especially when a single digital face can be used globally across markets without the logistical costs of a physical shoot.
The obvious benefit to these introductions is the cost on brands - there is no hair-stylist, makeup artist, catering, or anything else required for an AI model.
If thoughtfully trained, AI can actually broaden representation. Forward-thinking developers are already working on datasets that prioritize a spectrum of skin tones, hair textures, body types, abilities, and gender expressions.
Imagine an online shopping experience where consumers can see a garment modeled on someone who looks like them—not just as a digital gimmick, but as a new standard of accessibility. AI can offer personalized fashion previews for people of all sizes and backgrounds, reducing reliance on one-size-fits-all imagery.
The potential is powerful. But potential alone isn’t progress—it takes intent. As the use of AI evolves it would be exciting to accommodate differently-abled representation, as there are virtually no cases of AI-generated models with physical disabilities in brand campaigns, while there are cases of differently-abled humans modelling in them.
One major challenge for AI is competing with well-known models and influencers, like Gisele Bundchen, Naomi Campbell, Gigi Hadid, and Kendall Jenner. Often brands want to align with these models because of their social impact and it would likely be hard to replace that with an AI-generated image, and likely expensive to recreate their likeness with technology.
In the end, the future is still unwritten. The future of fashion shouldn’t be a digital illusion of diversity—it should be a continuation of real, visible, meaningful inclusion. That future is still within reach—if the industry chooses to shape it that way.
Cover Image Source: Oleg Ivanov