As artificial intelligence slowly but surely infiltrates fashion production, it’s also beginning to shake up the way trend forecasters pinpoint emerging styles.
A number of data programs are changing the way fashion forecasters and editors examine trends. Though retail analytics firm Edited has used machine learning for the past eight years to identify common themes involving color at runway shows, it is now enhancing its capabilities to assess more complicated areas, including silhouettes, textures and prints.
Meanwhile, companies like Tagwalk — which has been dubbed the first online search engine for fashion shows and touts prominent retail investor Carmen Busquets as a partner — are using the technology to change how the fashion industry analyzes runway looks. In addition to amassing every style from all the major shows on its platform, Tagwalk profits by selling data reports compiled using algorithms on the site, specifically tailored for brands based on areas of interest.
While the futuristic perception of AI evokes images of robots taking over jobs, Edited senior retail analyst Katie Smith said this is not the case.
“Machine learning and AI is augmenting, rather than replacing jobs in the apparel retail industry,” she said. “Historically, the industry has lacked the specific data and insights available to make smart and strategic retail decisions. Previously, designers, buyers, merchandisers and style editors would need to go into physical stores [or runway shows] to understand the latest trends, or rely on guesswork to anticipate what’s next.”
Edited’s latest tool employs a method called deep learning, a process that uses a series of prototypes that train computers to better identify specific features and nuances in images that transcend just color. It helps the team extract comprehensive insights of runway looks more so than previous algorithmic strategies, ultimately behaving “more like the human brain, spotting patterns and subtle features in stimulus,” Smith wrote in her Edited report.
An example of the prototypes used to track trends for London Fashion Week, courtesy of Edited
Carla Buzasi — global chief content officer at WGSN, one of the largest fashion trend forecasting companies in the world — said the role of the forecaster will not become obsolete. While the rise of AI is changing the speed with which WGSN can examine data, she said the company will always rely heavily on human resources to make definitive declarations on trends.
For WGSN, the key is these proclamations pair data with real world considerations and experiences that exist beyond the realm of bots.
“AI is changing the world, and trend forecasting is not immune to this change,” Buzasi said. ”The use of big data is already enabling us to forecast trends more accurately and more quickly, but none of this is possible without the creative force of our team of experts. At WGSN we strongly believe in the dual power of data and people to forecast trends.”
Buzasi added WGSN has invested heavily in its own machine learning tools in the last several years, creating products like WGSN Barometer, which measures consumer perception across brands and retailers using 100 different metrics like social metrics and sales. These efforts are paired with its staff of 250 employees in 14 countries, who routinely travel around the world to understand how data findings are playing out in daily fashion choices.
“Technology needs training by experts, and those same experts also need to judge and revise the predictions that AI might make,” she said. “We believe data is great at the ‘what’. People provide the ‘why’. Bringing these two together is the future of trend forecasting.”
Ashley Paintsil, fashion journalism professor at University of Delaware and former editorial director of FashInvest, said she also isn’t concerned that AI will take the place of editorial roles. Echoing Buzasi, she said only humans can understand how style trends exist within the real world and make accurate predictions for the future.
“At the end of the day, trend forecasting relies on data,” she said. “Having that data is useful, but there are human trend forecasters for a reason. A machine can’t go out and see four people wearing a black skirt on the street and anticipate that’s going to be a trend. A human being can do that.”
On the other hand, Smith said an important benefit of machine learning is uncovering or tracking trends that may go overlooked.
“The value of AI is that it can spot potential advantages that may not be immediately obvious to companies,” she said. “We see a world where ultimately, fashion and retail is transformed to be better, faster, more efficient and more profitable, because technologies are helping consumers see the right products at the right time and prices, appealing to their individual preferences and desires.”
Photo courtesy of Creative Applications Network