Skip to navigationSkip to content
Designs by a human.
Reuters/Thomas Peter
Can a machine ever know more about your style than you?
ALGORITHMIC AESTHETIC

In the future we’ll all have personal stylists designing us customized clothes

By Ephrat Livni

Though not typically known for their fashion sense, scientists do have style: Creative engineers are pioneering an AI that will be able to design clothes suited to your unique tastes.

Algorithms already inform many of the clothing recommendations we see online, and retailers like Stitch Fix are working on clothing designs generated in part by AI suggestions. Now, a team of technologists from the University of California-San Diego and Adobe Research are working on machines with brains that can create custom clothing designs based on a wide range of data, including buyer preferences, according to a Nov. 7 study (pdf) in ArXiv.

The main challenge, the researchers noted, is that most recommendation engines are based on shoppers’ purchases, and don’t address things like changing trends or personal stylization. To combat those obstacles, they needed an AI that could “judge” designs based on certain characteristics, such as color, texture, or shape, and also factor in individual purchase history. Because when it comes to being a fashion icon, you won’t succeed by wearing the clothes that were bought most often by strangers.

UC San Diego
Not quite like having Stella McCartney designing for you. But close enough for jazz?

When data on visual characteristics was combined with individual purchase history, the researchers were able to provide bespoke user recommendations, either from existing designs, or by generating new designs, altered to fit individual style.

The influence of artificial intelligence on fashion is still in its early stages, and the researchers concede that visual characteristics like color and texture are not in and of themselves “the subtle characteristics of ‘fashion.'” But while we’re still a ways away from having robot personal stylists, bespoke clothes for the masses could be possible someday soon.