The General Services Administration is carrying out a study to test and validate potential racial bias in facial recognition technology systems.
As part of the inquiry, the agency will study the equity of certain remote identity-proofing technologies that the American public may use when accessing federal benefits. GSA plans to release a report with the results from the study in a peer-reviewed publication in 2024.
The study comes after major concerns over equity were one of several concerns raised by a May watchdog report highlighting failures in the development of GSA’s Login.gov facial recognition platform. According to that report, the concerns were a key reason for delaying the system rollout to certain government customers.
In a note on LinkedIn announcing the study, Commissioner of the Federal Acquisition Service at GSA Sonny Hashmi said: “This is an important study and initiative to test and validate facial recognition and matching algorithms and technology to identify barriers across demographic lines.”
“The results will not only inform government strategy moving forward, but will also lower barriers for more Americans when they interact with their government digitally,” he added.
In recent years, government watchdogs and independent nonprofits have long raised concerns about the potential for facial recognition technology used by federal agencies to encode racial bias. In 2019, the GAO called on law enforcement agencies to further study the impact of facial recognition technology on equity.
A study by the National Institute of Standards and Technology, which was published in late 2019, found that facial recognition systems falsely identified African-American and Asian faces 10 times to 100 times more than Caucasian faces. According to that study, among a database of photos used by law enforcement agencies in the U.S., the highest error rates came in identifying Native Americans.
GSA is seeking prospective study participants, and anyone who takes part in the study will receive a $25 gift card.