Skip to main content

Rise of the AI fashion police

Rise of the AI fashion police

/

Can bots really help you dress better?

Share this story

If you buy something from a Verge link, Vox Media may earn a commission. See our ethics statement.

Amazon

There are two kinds of people in this world: those who want to know what the back of their hair looks like, and those who firmly do not.

That first group is likely to be thrilled by Echo Look, the “style assistant” Amazon announced last week. It’s basically an Echo combined with a hands-free camera controlled by your voice (“Alexa, take a picture”), and it’s being marketed — at least to start — as a way to capture outfit photos. The Look can also record video, so users can see what they look like from every angle, which is either a godsend or horrifying, depending on which side of the aisle you sit.

Amazon’s language around the Look makes Alexa sound like a fashion-savvy buddy. “Alexa helps with thousands of things, and now she can help you look your best,” declares the promo video accompanying the Look’s announcement. But it’s not hard to imagine an ulterior motive. After all, who has more to gain from well-lensed, head-to-toe outfit photos — you or Amazon? Fashion is one of the company’s fastest-growing categories; what could be more useful than a vast bank of user-submitted photos to learn the tastes and existing wardrobes of its customers, and where there might be holes to fill?

Who has more to gain from well-lensed, head-to-toe outfit photos — you or Amazon?

Just weeks ago, the company was granted US government approval on a manufacturing patent for a fully machine-operated apparel factory with the capability to produce clothing “on demand.” If built, the factory could feed Amazon’s private label brands, a significant focus for the company. There are currently 40 jobs open for its private label business, indicating expansion across activewear, men’s, and kids’ clothing.

Despite the possible gains for the company’s apparel-slinging business and valid privacy concerns, there is some value Amazon can bring to the process of getting dressed. Echo Look comes with a feature called Style Check, which can offer “a second opinion” on your outfit. To use it, you submit two of your own outfit photos — captured with or without the Look — using a corresponding app. Style Check then spits back an opinion on which is “better” (subjective!) based on “current trends and what flatters you,” the company explains, as determined by “machine learning algorithms with advice from fashion specialists.”

Style Check is not dissimilar to Amazon’s existing Outfit Compare feature, which Prime members will find nestled under programs and features within the iOS app. The big difference between the two is the amount of human input. “The responses in Style Check are a hybrid of machine learning and fashion specialists, while Outfit Compare provides the opinion of a single fashion specialist,” Michelle Taylerson, senior PR manager for Amazon Devices, tells The Verge. “Style Check responses are provided as a percentage, offering a detailed look at how much the selected outfit is preferred, as opposed to the three levels of differentiation in Outfit Compare (It was a close call! / We like this better! / Definitely pick this one!).”

The concept — make getting dressed easier with the help of a digital authority — isn’t new. In 1995’s Clueless, Cher Horowitz’s expansive closet is digitally cataloged so she can flip through tops and bottoms on a touchscreen desktop computer (in 1995!) to create her own looks. If the outfit doesn’t work, the computer then declares it a “MIS-MATCH!”; if it works, the computer shows how it would look on a perfectly modeled image of Cher. Twenty-two years later, why we don’t have this is beyond me.

Various tech and fashion companies have taken stabs at similar concepts over the years, but they’ve all fallen short, offering just part of the experience: wardrobe cataloging, outfit suggestions, or 3D modeling.

Gap announced a try-on app called the DressingRoom at CES this year, currently only available on Google Tango-enabled devices. Users can only pick one of five body types to try on Gap clothes, so it’s pretty limiting. H&M did something similar way back in 2008, which it has since abandoned. Closet-organizing apps are a whole other, equally imperfect beast. Stylebook is an app that works with your existing closet to plan and save outfits, but there is no feature that weighs in on your looks. GlamOutfit does the same, as do many, many others. Then there’s My Virtual Model, which has been kicking around since 2011 and can create a pretty convincing digital avatar of yourself. With MVM, you can try on generic clothes (like a red tank top or wide-leg trousers) as well as clothes from selected partner retailers, but there is no way to integrate your existing wardrobe or get feedback on your looks.

The best existing style AI advice comes from Epytom, a stylist bot that lives on Facebook Messenger, boasting 550,000 daily users. Based on a capsule wardrobe of 40 common clothing items, the bot shoots out daily outfit ideas for users, taking into account gender, age, current weather conditions, and a user-selected style profile. For example, a recent suggestion for me (female, 28 years old, 90026, “artsy”): “It is sky is clear, a high of 76F and a low of 58F, and this Friday is looking great on you!😍 Let’s ease into the awesomeness with Little Black Dress and Crew-Neck Tee!🖌” The bones of my outfit are determined, and a follow-up message provides further detail: “Strappy slip dress and a white tee are the best for this bold combo — you'll make a loud statement with the contrast! For the best proportions, choose a chic midi dress hit mid-shin and pair it with flats,” followed by three photo examples of the outfit, sourced from style bloggers.

You can’t submit photos of your own outfits, and the language is clunky, but it does tell you what to wear. “Even among all the variability in fashion, there are style formulas that simply work,Epytom content editor-in-chief Marianna Milkis-Edwards tells The Verge. She explains that Epytom has analyzed outfits from style icons of the past, like Audrey Hepburn, as well as today’s top tastemakers. “I believe the Echo Look was born out of popular requests; ‘What should I wear today?’ definitely being one of them,” she says. “Right now, Echo is gathering user data and helps you choose a better option. Epytom works a step prior, helping you put together an outfit to show off to Alexa.”

“Even among all the variability in fashion, there are style formulas that simply work.

Amazon’s existing Outfit Compare is the cleanest, most streamlined, and personalized “stylist” on the market, but its judgement relies entirely on old-fashioned human input. And why should you trust an Amazon employee’s opinion of your look? The company explains that its fashion specialists are well qualified, with “backgrounds in the fashion retail, editorial, styling, and creative fields,” who have “honed their expertise through ongoing training focusing on personal style, seasonal trends, and more.” But that doesn’t change the fact that taste is subjective. And adding an algorithm to the mix, like Look does, probably won’t help matters.

At their best, these digital style companions offer little more than a jumping-off point when it comes to getting dressed or developing personal style. At their worst, these “style assistants” perpetuate a dated binary of a “right” and “wrong” way to dress. The reality is that the slow-to-change fashion industry is finally beginning to embrace diversity across race, size, and gender. The big message is “be unapologetically yourself” (finally!), with decreasing emphasis on seasonal trends and a complete abolishment of “the rules.”

No doubt, for many people, getting dressed is a stressful and confusing time-suck. “We love to think about fashion as an unbounded self-expression, but you’ll be surprised at how many people just want an outfit that a) suits them, b) is coordinated and not boring, and c) doesn’t require them to spend hours in front of the mirror,” Milkis-Edwards says. The promise of an authority is attractive: I don’t care about fashion, but I need to wear clothes in public, and I’d prefer not to look like a damn fool.

For this sect, perhaps Style Check or Outfit Compare will be an on-ramp to developing personal style, or at least inspiring the confidence to get dressed without it being This Whole Thing. Say what you will about the frivolity of clothes, but your outward appearance has gravity, communicating who you are — or aren’t — before you have a chance to open your mouth. Clothes, accessories, hair style, and makeup can be powerful tools. Outside of nudist colonies, everyone has to fuss with these elements (despite their interest level), so you might as well use them to your advantage. If Amazon’s new tools are what helps you get there, cool. But you could always text a mirror pic to a trustworthy friend and ask what they think.