Osagie K. Obasogie in the Boston Review:
For several years now, a quiet revolution has been underway in consumer electronics. Gadgets that are a part of our everyday lives have learned to see. Relying on optical devices and software that can detect faces and track body motions, cameras, gaming systems, phones, and other tools have gained access to the mechanisms of visions and recognition that were once considered the unique province of sentient beings.
But some of these technologies have a problem: they have a hard time seeing people of color.
In December 2009 Wanda Zamen and Desi Cryer, two employees at a camping supply store in Waller, Texas, noticed something peculiar about an HP computer at the shop. The computer featured a digital camera that detected and tracked human faces. The system had no problem identifying and following Wanda, who is white, but it could not do the same for Desi, who is black. He demonstrates the glitch in a YouTube video that has been viewed almost 3 million times. “As you can see, the camera is panning to show Wanda’s face. It’s following her around. But as soon as my blackness enters the frame, . . . [the camera] stops,” he says.
A similar bug was found in the Nikon Coolpix S630 digital camera. Designed to overcome the timeless challenge of a blinking subject, the camera detects faces and alerts the photographer when it senses closed eyelids. But when Joz Wang, a young Taiwanese woman from Los Angeles, tried to take pictures of her family, the camera kept showing the same error message: “Did someone blink?” To which Wang responded on her blog, “No, I did not blink. I’m just Asian.”
And Microsoft’s popular Xbox 360 Kinect video game system, which uses facial and body-motion detection to enable interaction, has also had trouble recognizing nonwhite faces.
All of these technological errors are correctable, yet the question remains: Why do they happen at all?