A couple of years ago, seven to be exact, the only La La Land everyone knew, including myself, was the film starring Ryan Gosling and Emma Stone. A few years down the line here is a Lalaland.ai that is making enough noise to overpower the effect of the namesake movie.
This is not only because it has the ‘AI’ tag attached to it but it is also because Lalaland.ai utilises a highly developed artificial intelligence to help fashion companies and stores generate incredibly lifelike models that represent various body shapes, ages, and skin tones. The reason why we are talking about it is because Levi Strauss and Co: the famous clothing company was subjected to backlash during the first quarter of 2023, after the announcement about their plan to use AI to generate models of more diverse body types and skin tones to expand representation of its products’ models.
The clothing company released a statement in March 2023 where they said that they will join hands with Lalaland.ai in order to “supplement human models” in an effort to “create a more inclusive, personal and sustainable shopping experience for fashion brands, retailers and customers”. However, many critics were confused as to why Levi’s would not use real models to do so rather than generate fake ones in the name of diversity. They were called ‘lazy’, ‘problematic’ and ‘racist’ because of this. Levi’s of course responded to the backlash and attempted to clarify the doubt by saying they would supplement human models and not replace them. However the uncertainty that is woven around this decision made not only by Levi’s but by many fashion brands hangs heavy in the air.
CGI
Computer Graphic Imagery? More like Computer Generated Inclusivity. This is a digital era and that is the plain, hard truth. Therefore, as surprised as we are to hear that the ramp will have an entire array of models that are AI generated, we cannot really say that we did not see this coming. Using AI, the cutting (in all senses of the word) edge technology, in every domain has now become the norm. And it is difficult to swim upstream when the current is determined by artificial intelligence. Therefore, it is not shock inducing to see clothing brands embrace AI models. However, many say that it is disheartening.
AI fashion models are computer-generated representations of models that are created using artificial intelligence. They serve a similar purpose as human models by displaying clothing and accessories for fashion brands. The unique feature of these AI models is that they can be tailored to exhibit different body shapes, ages, sizes, and skin tones, providing a diverse array of appearances to showcase fashion products.
Unlike static digital mannequins, these AI models are dynamic and versatile. They can change their looks, styles, and backgrounds instantly with a simple command. It is said that the purpose behind these AI models is to offer innovative ways for presenting fashion collections, enabling businesses to quickly modify, experiment with, and personalise various styles as needed.
While this may have its advantages such as cost reductions, customizability, diversity and inclusivity, sustainability and better shopping experiences, the advent of AI models brings its fair share of challenges. Ones that may very well override the pros.
Many individuals in the world are fighting with all their might so that they can discard the narratives that champion unrealistic beauty standards. However, creating digital models will undo years worth of effort. Especially because they come into the picture with a level of perfection that is beyond what humans can achieve. This pursuit of ‘perfection’ will promote unrealistic ideals of beauty and worsen the challenges people face with their body image.
For instance AI models that are used by clothing companies in China, on the pretext of being ‘inclusive’, are creating ones that are slim, busty and dressed only in a black lace bra and stockings. Shu Xinrou in an article titled “In China, AI-Generated Fashion Models Are Hugely Popular-and Sexist” explains how AI-generated images such as the ones explained above have taken China’s fashion industry by storm in the past few months. It is horrifying to see that these AI models “conform to a certain-highly sexualised-ideal of female beauty: one that prizes tiny waists, wide-hips, and visible cleavage”. While it is clear as day that these images are catered exclusively for the ‘male gaze’, women are horrified about the way in which they are portrayed and as to how this will reinforce toxic beauty standards.
Additionally the shift from human models to AI counterparts could also have economic consequences for the fashion industry. There might be implications related to job displacement and changes in the traditional fashion model ecosystem.
AI models will definitely be caught on the ethical radar where they will be swamped with discussions regarding issues such as consent and the potential misuse of digitally created images. Furthermore, real human models contribute a sense of genuine connection and authenticity to brands. Consumers who value these personal and authentic experiences might not be as receptive to AI models, as they might feel detached or disconnected from the brand’s message. Even from a business and marketing perspective, this may seem like a foul idea.
A paradox of diversity
There is a significant outburst ensuing the incident with Levi’s where many took to twitter to voice their opinion. The furore around the initiative is caused mainly because creating an avatar of a person of colour- rather than opting for a real person- by no means appears like a solution to address the lack of diversity in the realm of fashion. It is mostly interpreted as an attempt to increase profit from the ‘appearance of diversity’ without actually having to invest in it. The use of AI models creates a controversy where questions such as who is profiting, which part of the representation is real and which of it is a hoax and what is the intention of those handling the technology are raised.
The controversy does not end there, many companies that are involved in creating AI models often forget that diversity does not only mean different skin colours and body types but it is also about persons with disabilities. Stephanie Thomas, also known as @disabilityfashionstylist, points out that AI cannot truly promote diversity when it is inherently ableist.
Stephanie Thomas is the founder of Cur8able and a stylist with experience in major adaptive fashion campaigns for brands like Nike, Zappos, and Kohl’s. She explains that this technology is predominantly developed by individuals without disabilities, who might view disability as a problem needing to be fixed or cured. Thomas emphasises that being a true advocate for diversity entails a comprehensive understanding of the various demographics being served. This also involves educating sales associates on how to dress and cater to people with disabilities.
While there is a firestorm of criticisms that has the concept of AI models at gunpoint, there are people who are of the belief that this can be used in an unproblematic and a good manner. Because despite the fact that companies like Levi’s have the money and resources to hire diverse models, some fashion stores may crash under the weight of the cost that should be incurred to not only hire diverse models but the process (such as makeup and photography) that ensues. Therefore, even though people have come down hard on big brands for being miserly and looking for makeshift solutions, they have not really scrutinised small scale clothing shops that barely make do.
Nevertheless, AI models are problematic, for more reasons than one.
Primarily because there is not enough diversified data that can be inputted to the algorithms for an unbiased outcome and this will always remain an issue. Consequently, quick-fixes like these would only undermine the gravity of the lack of diversity making it look like a lesser problem than it actually is. Moreover, with the acceleration of AI activations, brands that desire to come out on the top will need to have an intimate grasp of what their audience really wants. However, there is always a gulf between what consumers want and what the companies assume they want. And this is what is fundamentally wrong in almost all industries and the concept of AI models is no exception.
(Sandunlekha Ekanayake)