The fashion industry asks AI to predict new trends: NPR

Predicting what counts in fashion was once an elite sport, the province of those who attend these major parades in New York, Milan, London and – at the moment – Paris, like the publishers of influential magazines. (A 2006 line The Devil Porte PradaAbout Miranda Priestly, a tyrannical fashion editor, transmitted the idea succinctly: “You still don’t get it? His opinion is the only one who counts.”)

While elite opinions continue to gain weight, the trend prediction game has developed enormously in the last decade. Platforms like Tiktok, Instagram and Pinterest redefine the way the trends are spreading. According to a report by the Launchmetrics data company, more than 40% of world consumers bought clothes and accessories at least three times via social media in 2024.

“There is more information available, certainly,” said Amy Sullivan, vice-president of the purchase and private brands at the online fashion retailer and the style point of style. “And I think trends because of social media move more quickly.”

With so much floating data, players in the fashion industry are based on AI in order to remain competitive.

Sullivan said the AI ​​recently helped his team decide to go with a red or blue striped shirt for next spring.

“In the past, to answer this question, you make a cash decision without really looking at it, or you ask for samples of vendors abroad that could take weeks and cost a lot of money,” said Sullivan. “But when it happened a few weeks ago, we just put it in AI and you can see it in a complete body image and make the right decision.”

They went with blue.

How AI is used to make predictions

AI algorithms help the fashion industry on a number of fronts, including the creation of personalized customer experiences, such as customers virtually “try” clothing, supply chain management and image generation. Its use to determine what is likely to steal racks next season implies the mining of many different places.

“Our models follow everything, from parades to social media,” said Noémie Voyer, fashion expertise in Heuritech. The company based in Paris works with brands such as New Balance, Skims and Prada. “We have a massive scale of data that allows our AI to detect the first signals of trends – sometimes months before becoming visible on the consumer market.”

VOYER said that his business algorithms predicted a bunch of emerging trends for next year. Examples include dotted prints, flat sandal and yellow color. These trends all presented themselves on the tracks of this year’s fashion weeks. Finally, they will appear in Target and H&M.

But see and the other sources with which NPR have spoken for this story all agree that AI cannot do the prediction of fashion.

“While our AI is extremely sophisticated, the human [aspect] Still remains essential, “said Voyer.

Human expertise is always necessary

“We can use AI to overeat what we do,” said Francesca Muston, director of forecasts for the Global Consumer Trend Prévade Company, WGSN, who counts Levi’s and Coach among her customers. “But you really have to start putting a lot of rigor and a lot of processes in what you do.”

Muston said AI is excellent to effectively predict things like the quantity of a popular article that a retailer should store. But she said that human experts from her business should ensure that the information provided by AI does not lead to bad conclusions.

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

Muston said it was one thing for an entertaining fashion article on social networks to get a lot of commitment. But is it something else to go out and buy the article in the spotlight: “Will they in fact disappear and spend money and bring it to the bar after work?”