Is AI going to be your next personal stylist?
Apparel seller Stitch Fix recently introduced a coral, sleeveless blouse with a split neckline—and an unusual creative provenance. It was one of three new tops designed with the help of artificial intelligence. The San Francisco-based e-commerce company, which sends customers boxes of preselected outfits, is leveraging computers to analyze purchasing behavior and learn what elements of style are popular. The software then recombines well-liked sleeve types, cuts and prints into new looks to maximize the odds a client loves the resulting style. The three tops sold out as part of preselected boxes last year, according to the six-year-old company, and in February, it started selling nine more items designed with the help of computers, including dresses and tops. It plans to sell more than two dozen others by the end of the year.
The “hybrid designs,” as they are known inside Stitch Fix, are part of a movement in the tech industry to develop software that can be creative, and produce content such as songs, logos, video games, clothing and special effects. The field of computational creativity dates back decades but is flourishing thanks to advances in machine learning, plus increased access to data and computing power. Alphabet Inc.’s Google, Adobe Systems Inc., Microsoft Corp., and Sony Corp. have active research projects related to computational creativity. Some, like Adobe, have spent millions in this space. Tech companies and researchers hope that teaching computers to be creative could lead to more powerful AI systems. Long term, the results could improve processes that require complex analysis, such as computer-vision systems in self-driving cars, according to machine-learning experts. And some companies, like Stitch Fix and Adobe, are already using such software to produce products.
One primary goal for tech firms is to create so-called general artificial intelligence—machines that excel at multiple tasks. Currently, AI systems are typically good at only one thing, like categorizing objects, and training the systems can require extensive help from humans. Today’s smart systems also aren’t very good at dealing with unpredictable situations, according to machine-learning experts. To get machines to learn on their own, some companies are employing what’s known as “adversarial training,” which pits two pieces of software against each other. The Facebook AI Research lab recently used the technique—developed at the University of Montreal—to make computer-generated images of churches and faces, among others. Others have since used it to create nearly photo-realistic images of ants, birds, monasteries and volcanoes. During adversarial training, one network tries to create images the other network can’t tell were dreamed up by a computer. From their interactions, the generator learns to create images on its own that can pass for real-world pictures and the other network figures out what’s real—and what’s fake.
The fashion industry is no stranger to fast cycles of learning. One of the great benefits Stitch Fix sees from collecting and analyzing so much data is an ability to predict trends. For example, the company’s engineers are developing machine learning classifiers to find trends by using the simple yes-or-no decision that a client makes when they buy an item or send it back. From this seemingly simple data, the team has been able to uncover which trends change with the seasons and which fashions are going out of style.
Another benefit of all this data is research aimed at developing a computer vision system that can interpret style and extract a kind of style measurement from images of clothes. The system itself would undergo unsupervised learning, taking in a huge number of images and then extracting patterns or features and deciding what kinds of styles are similar to each other. This “auto-styler” could be used to automatically sort inventory and improve selections for customers.
In addition to developing an algorithmic trend-spotter and an auto-styler, Stitch Fix is developing brand new styles — fashions born entirely from data as mentioned in the first paragraph. The company calls them “frankenstyles.” These new styles are created from a “genetic algorithm,” modeled after the process of natural selection in biological evolution. The company’s genetic algorithm starts with existing styles that are randomly modified over the course of many simulated “generations.” Over time, a sleeve style from one garment and a color or pattern from another, for instance, “evolve” into a whole new shirt.
We’re only at the beginning of the era of artificial intelligence. Some upheaval is to be expected. But we are starting to see how AI can change industries, improve productivity, and even benefit a new generation of employees.