Facial recognition is a fast-developing technology that is becoming pervasive in our lives—just think of the iPhone’s Face ID. By capturing various “points” on people’s faces, facial recognition can not only identify what we look like, it can also identify our moods. The use of facial recognition has always been a subject of debate. People worry about the technology’s accuracy and its potential security risks, and in the wake of civil rights protests in the US, there is also widespread concern that police are abusing facial recognition to identify protesters.
Yet as researchers continue to improve facial recognition, this technology is gradually expanding from airports and surveillance cameras to more mundane activities, like eating out at restaurants. Self-ordering kiosks equipped with facial recognition, which tech companies like GRUBBRR are developing, can identify customers’ race, gender and age and make suggestions for their orders based on these results.
A question is inevitably raised: if race, gender and age are the categories kiosks use to make recommendations, won’t they propagate stereotypes? There are many long-standing stereotypes in America revolving food. Food commercials often craft a narrative that salad is for women while steak is for men, and sometimes food stereotypes can carry dark connotations. For instance, many cartoons used to depict African Americans as always eating “fried chicken and watermelon,” two foods eaten with our hands, to undermine them as dirty and “childlike” and thereby marginalize them further. This association obviously foments racial prejudice.
Even though in a restaurant customers won’t know what suggestions other customers receive at their respective kiosks, they may still get offended if they feel profiled by a machine based on race, gender or age. Maybe we can detect what specific groups like to eat by examining data, but we must remember trends cannot always be applied to all individuals within that group. A teenager religiously devoted to healthy eating might get really frustrated if she constantly receives recommendations for salty, oily foods at a self-ordering kiosk. For restaurant owners who want to make customers feel they are receiving customized service — in fact, this is why a lot of restaurants are turning towards self-ordering kiosks in the first place — these assumptions are counterintuitive. No one wants an order suggestion based on a generalization.
In response to these issues, researchers and developers working on self-ordering kiosks with facial recognition might consider relying more on customers’ moods to make order suggestions. They can also ensure customers know recommendations are not based on stereotypes by making kiosks display phrases like, “You look a little down today, how about a banana pudding to cheer you up?” This will also show customers that the restaurant cares about them personally.
Though people still have different preferences for food even when they have similar moods, making food suggestions based on mood is not offensive. After all, the common emotions we share bond us as a species, regardless of race, gender or age.
