Social Media and AI Can Help Boost Fashion Forecasting社交媒体和人工智能助推时尚预测

A combination of AI and creativity translates our cultural values into colors and shapes.

人工智能和创造力的组合将我们的文化理念转化为色彩和样式。

Sometimes it feels like new fashion trends just pop up out of nowhere. In reality， though， these fads are usually the product of months or even years of careful observation and planning by behind-the-scenes actors： trend forecasters.

有时，好像新的潮流趋势不知怎么突然就冒出来了。而实际上，它们通常是幕后参与者（即潮流趋势预测者们）经过数月甚至数年仔细观察和策划的产物。

These craze-setters take note of fashion shows and celebrity looks， but they also collect data on politics， entertainment， the environment， technology， and consumer behavior. Such big-picture cultural observations are a jumping-off point1 to infer what colors， silhouettes， and fabrics will be the macro trends in fashion up to two years in advance.

这些热潮引领人士关注时装秀和名人造型，也收集政治、娱乐、环境、技术和消费者行为的数据。这种宏观的文化观察是一个出发点，可以提前推测出哪种颜色、样式和面料将成为未来两年的大规模流行趋势。

Where do fads come from？

潮流从何而来？

“Fashion is always a reaction to the environment，” says Carrera Kurnik， lead editor of Culture and Consumer Insights at Fashion Snoops， a trend forecasting company. “Economic climates， emotional climates， political climates， and all these different things that are happening around us are affecting how we dress， how we want to be viewed， and how we show others who we are.”

“时尚一直是对环境的反应。”时尚趋势预测公司Fashion Snoops文化与消费者洞察专栏的主编卡雷拉·库尔尼克说道，“经济形势、情绪氛围、政治气侯，还有我们周围所有这些正在发生的各种事情都影响着我们的穿着、我们希望怎样被看待，以及我们如何向别人展示自我。”

Trend forecasters begin with observations of changes in any of these climates. According to Francesca Muston， Vice President of Fashion at the forecasting company WGSN， the real analysis begins after a series of similar observations occur. For example， WGSN forecasters found several years ago that millennials and Gen-Z2 were being priced out of big cities， and also that freelance and gig work3 was booming while delivery services were increasing. On top of all that， overall anxiety levels seemed to be rising. That all led to the large-scale insight that people would be spending more time at home， says Muston. Looking back at historical data， Muston adds， they inferred that this would mean more folks would desire comfortable clothing.

要做好潮流趋势预测者，首先要观察上述各种环境中的变化。据潮流预测公司WGSN的时尚副总裁弗朗西斯卡·马斯顿说，一系列类似的观察完成后，真正的分析才开始。例如，几年前，WGSN的预测人士发现千禧一代和Z世代因负担不起高昂生活成本而离开大城市，而且，在递送服务增加的同时，自由职业和零工工作也在蓬勃发展。此外，总体的焦虑水平似乎在上升。马斯顿称，这些造成了一个普遍性观点，即人们会花更多时间待在家里。马斯顿补充道，回顾历史数据，他们推断这意味着更多人会想要拥有舒适的衣服。

After identifying a cultural trend， analysts work to translate it into a corresponding shift in fashion. Forecasting companies have access to huge amounts of historical data—runway photography， fashion magazine archives， and their own research—on top of all the information constantly flowing out of potential consumers on social media. Forecasters search for previous instances of similar trends， associate values like “comfort” with colors， fabrics and silhouettes， then settle on recommendations that their clients use to decide what to design.

确定文化趋势后，分析师们会努力在时尚领域做出相应调整。除了通过社交媒体从潜在消费者那里获得源源不断的信息外，预测公司还能获得大量历史数据，其中包括时装秀照片、时尚杂志档案以及自己的研究资料。预测者们搜寻之前类似流行趋势的实例，将像“舒适”这样的观念与颜色、面料和样式联系起来，然后向客户提出建议，供他们设计决策时参考。

In the past， new fads appeared almost exclusively through a “top-down” method， in which fashion designers release their runway collections and those new styles catch on slowly. For forecasters， that analysis is fairly straightforward; it takes about two years for couture elements to make it into retail stores. The other way is “bottom-up，” which is where trends emerge organ-ically in culture—like when the cool kids in a particular city start wearing a new style that then catches on in other social groups and geographic locations.

过去，新潮流几乎完全是通过“自上而下”的方式出现的，时尚设计师发布自己的时装系列，而后这些新款逐渐流行起来。对于预测人士，这类分析相当简单，高级时装元素要花大概两年时间才能进入零售店。另一种方式是“自下而上”的，即流行趋势在文化中自然而然地出现。比如某个城市的时髦少年开始穿着一种新款式，随后这种风格在其他社会群体和地区逐渐流行起来。

Nearly all trends used to trickle down4 from the elite ranks of the fashion industry， but social media and influencer culture have changed that. “The internet has really shaken up the almighty trickle-down，” says Kurnik.

过去几乎所有的潮流都是从时尚界的精英阶层逐层扩散的，但社交媒体和网红文化已经改变了这一情况。库尔尼克说道：“互联网真的颠覆了那种居高临下的涓滴模式。”

The AI trendspotters

人工智能时尚发掘者

Social media isn’t the only thing that’s drastically changed the way that trend forecasters make their predictions; many companies now use artificial intelligence programs to hunt down new trends. Fashion Snoops， for example， uses AI to scrape the internet for buzzwords and novel slang with the potential to develop into something chic.

社交媒体并不是唯一彻底改变预测者预测方式的东西；许多公司如今使用人工智能程序来捕捉新潮流。例如，Fashion Snoops利用人工智能在网上搜索有可能发展成时尚因素的流行语和新奇俚语。

IBM’s Watson5 AI can analyze hundreds of thousands of images from runway shows and generate insights about what colors and patterns retailers should be looking to stock for the upcoming season. The algorithm disregards irrele-vant data， like types of backgrounds and models’ skin tones. Then， it finds the prominent colors in each image and records them， eventually returning data about how often each color occurs. It can run a similar analysis for fabric patterns and for finding similarities between different runway shows. An individual fashion trend forecaster would never be able to analyze that amount of data in time for the next season， so using AI to do the grunt work frees up forecasters to look for upcoming trends in less conventional areas， like movies， TV， or even politics.

IBM的沃森AI能够分析数十万时装秀照片，并为零售商们下一季应该储备哪些颜色和样式提供见解。该算法忽略了无关数据，比如背景类型和模特的肤色。随后，它会找到每张照片的突出颜色并记录下来，最后返回每种颜色出现频率的数据。它能对面料图案进行类似的分析，并找出不同时装秀之间的相似之处。单靠一位潮流预测人员绝不可能在下一季到来前及时完成对如此大体量数据的分析。因此，利用人工智能完成繁重的工作，可以让预测人士腾出时间在非传统领域（如电影界、电视界，甚或政界）发掘即将到来的潮流趋势。