The fashion industry is asking AI to predict new trends : NPR

Having predicted what was important in fashion, it was once elite sports, the province of those who visit these large shows in the runway in New York, Milan, London and -Paris, like editors of influential magazines. (Line from 2006 The devil wears PradaAbout Miranda, a priest, a tyrannical editor of fashion, briefly conveys the idea: “You still do not understand, do you? Her opinion is the only thing that matters. “)

While elite opinions continue to wear weight, the game forecasting has expanded significantly over the past decade. Platforms, such as Tiktok, Instagram and Pinterest, over detecting how trends spread. According to report From the Launchmetrics, Data Company more than 40% of global consumers purchased clothes and accessories at least three times through social networks in 2024.

“Of course, there is more information,” said Amy Sullivan, vice -president for buying and private brands in the online fashion store and style service store Stech fixThe field “And I think that trends from social networks are moving faster.”

With such a large amount of data, the players of the fashion industry are based on AI in order to remain competitive.

Sullivan said that AI recently helped her team decide whether it was worth going with a red or blue shirt of the strip for the next spring.

“In the past, in order to answer this question, you either make a point decision without looking at it, or ask samples from suppliers abroad that can take weeks and cost a lot of money,” Sullivan said. “But when this happened a few weeks ago, we simply placed it in AI, and you can really see it in the full image on the body and make the right decision.”

They went with the blue.

How is II used for forecasting

Artificial intelligence algorithms help the fashion industry in a number of fronts, including the creation of personalized customer experience, for example, allow customers to “try on” clothes practically, management of supply chains and images generation. Its use when clarifying what can fly out of the rack in the next season, includes data on the extraction of minerals from different places.

“Our models monitor everything from the show of take -off strips to social networks,” said the war, a fashionable examination HeuritechThe Paris Company is working with brands such as New Balance, Skims and Prada. “We have a huge scale of data that allow our AI to detect early trends – sometimes a few months before they become visible in the main market.”

Wars said that her company's algorithms are successful Predued A bunch of new trends for the next year. Examples include dashed prints, sandal with flat and yellow. All these trends appeared on take -off -landing stripes at fashion weeks this year. In the end, they will appear in Target and H&M.

But Voyer and other NPR sources spoke for this story everyone agrees that AI cannot do the prediction of fashion on their own.

“While our AI is extremely complex, a person [aspect] Still remains necessary, ”said the War.

Human examination is still necessary

“We can use the AI ​​to overload what we are doing,” said Francesca Muston, director of forecasting Global Consumer Trend Company, WGSNwho considers Levy and coach among his customers. “But you must really start investing a lot of severity and a lot of processes in what you are doing.”

Mredred said that AI is great for the effective prediction of things, for example, which popular item should sell a retail seller. But she said that human experts of her company should provide information provided by artificial intelligence, does not lead to incorrect conclusions.

“When you look on the Internet on social networks, the trends can explode and feel huge,” Muston said. “But entertainment and commerce are two different things.”

Mredred said that this is one for an entertaining fashion post on social networks to get a lot of involvement. But this is completely different in order to go out and buy an object: “They will actually leave, spend some money and wear it in the bar after work?”