Some recent tests, conducted by The North Face with an innovative Expert Personal Shopper program, suggests that there is much to be learned when it comes to using artificial intelligence in retailing. In particular, the program, produced by a company called Fluid, seeks to learn exactly what consumers are looking for when they first click on to The North Face’s website or mobile channel—just like a sales clerk would do if they walked into a North Face store. By asking questions with progressively increasing detail, the program (powered by IBM’s famous Watson) selects several items that might meet a shoppers needs and presents them onscreen. The two-way communication seeks to feel natural and organic, rather than imposing selections on the shopper. But before it achieve this status or devise accurate recommendations, the program first must learn about all the products available. A sophisticated search assistant is little help if it can’t recognize the meaning of product descriptions to be able to match items with consumers’ wants. In addition, it needs to integrate external information that might be pertinent: For a consumer searching for a jacket to wear on a ski trip to Vermont in December, for example, the Fluid program needs to know what the average temperatures are at that time and place. But as these lessons get learned, this experiment might predict the future of retailing. The 50,000 customers who worked with a beta version of the software remained on the site for an average of 2 minutes longer than they did without the assistant. They also gave it high ratings for its functionality, and 75 percent of them noted that they would use it again.

Source: National Retail Foundation, January 18, 2016, https://medium.com/nrf-events/it-s-time-for-a-conversation-on-artificial-intelligence-30aea7d73f69#.tqi8g4jc9