A report published by the National Institute of Standards and Technology (NIST) is showing that facial recognition technology often performs unevenly based on a person’s race, gender, or age. NIST, which is a part of the Commerce Department, has been studying facial recognition for nearly two decades as it is increasingly being adopted by law enforcement, airports, and a variety of businesses. The study could now fundamentally change the way this type of technology is used.
For its research, NIST tested 189 algorithms of 99 companies, academic institutions and other developers that voluntarily submitted their technology for review. The algorithms form the central building blocks for most of the facial-recognition systems around the world. Those algorithms were run on about 18 million photos of more than 8 million people in FBI mugshots, visa application photos, and other government-held portrait images. The researchers said no photos were taken from social media, video surveillance or the open Internet.
The researchers tested both false negatives, in which the system fails to realize two identical faces are the same, as well as false positives, in which the system identifies two different faces as being the same. They found that many facial recognition technology systems misidentify people of color at a higher rate than white people. In one-to-one matching, which is generally used for verification, Asian and African American people were up to 100 times more likely to be misidentified than white men. In one-to-many matching, used by law enforcement to identify people of interest, faces of African American women resulted in the most false positives.
Middle-aged white men generally had from the highest accuracy rates, according to the study. Women were more likely to be falsely identified than men and Native Americans had the highest false-positive rate of all ethnicities. The elderly and children were more likely to be misidentified than those in other age groups. The results are alarming because facial recognition technology has become one of American law enforcement’s fastest-growing tools for identifying criminal suspects and witnesses.