Artificial intelligence (AI) is all around us, but emotion AI systems go further. They’re designed to detect and interpret a myriad of human emotions by gathering, analyzing, comparing and reading data points such as body language, gestures, brain waves, facial expressions and tone of voice. 

There are risks to AI systems interpreting human emotions. Research suggests that while sometimes people do smile when they’re happy, how people communicate the basic emotions varies greatly across cultures and situations. Furthermore, individual differences in personalities result in different ways of communicating, both verbally and non-verbally. 

"Such AI technology—computer vision on faces—easily exhibits ‘performance gaps’ by race. This is a red flag. Because this is such a prevalent problem for all AI that works on people’s faces, companies should be extremely cautious," says Johannes Himmelreich, assistant professor of public administration and international affairs and senior research associate in the Autonomous Systems Policy Institute.

Read more in the Disruptor article, "Emotion AI: Can artificial intelligence really read humans?"