China’s KFC Testing Facial Recognition Software For Suggestive Selling

December 28, 2016 by Dave Haynes

In a move that seems much more a gimmick than a boost of customer experience, the Chinese business unit of KFC is using video analytics at an interactive screen to suggest what customers might like to order, based on what they generally look like.

As reported in TechCrunch, and re-reported all over the place based on that, KFC is working with Baidu (China’s Google) on a single “smart” restaurant in Beijing that uses a camera to detect and scan the geometry of the face looking at the order screen, and use machine learning to then make a suggestion on what that person is likely to order, based on things like age, gender and facial expression.

This is a riff on the endlessly touted, rarely used idea of serving ads tailored to the general type of person watching. The truly interesting thing here is that this is NOT just anonymous face pattern detection (as is the case with the vast majority of the video analytics systems marketed for things like digital signage and digital out of home advertising). This is the real deal – facial recognition.

Reports TechCrunch:

“And the setup also has built-in recognition, so if you’re a return customer, it can ‘remember’ what you ordered before and suggest your past favorites. Of course, the idea of a fast food restaurant retaining my image for recognition purposes, tied to an actual order history, is more than a little unnerving.”

Privacy advocates in western countries would be raising all kinds of alarms about that. This is not anonymous data collection that helps build some general trends around buying patterns. This is a fast food company and a search engine giant taking your photo and storing it.

There’s also, as Quartz points out via a Wired magazine piece, the whole questions as to whether how people look has much or any bearing on what they want.