Uncovering ‘Coded Bias’

While working on a project that used facial recognition software, MIT Media Lab researcher Joy Buolamwini was startled to discover that the algorithm was unable to recognize her own face—until she put on a white mask.

Buolamwini, who is Ghanaian American, soon realized that much of the facial recognition software in use would often misidentify women and darker-skinned faces. Artificial intelligence (AI) programs—the basis for facial recognition technology and other software used for policing, advertising, hiring and financial services—generally have been trained to identify patterns based on data sets that skew light-skinned and male.

Her work on racial and gender biases in AI is the basis for “Coded Bias,” an award-winning documentary available online to the University of Maryland community starting tomorrow, with an online panel discussion of algorithmic biases scheduled for 4 p.m. Friday. 

The documentary and discussion are being presented by the Maryland Center for Women in Computing, part of the Iribe Initiative for Inclusion and Diversity in Computing, with additional support coming from UMD’s Student Entertainment Events and the College of Computer, Mathematical, and Natural Sciences.

We’re at a pivotal moment in the world of computing, said Jan Plane, a principal lecturer in computer science who leads both the center and the Iribe initiative.

“The workforce in technology continues to change, with more women and other underrepresented groups bringing diverse backgrounds and skills to projects,” she said. “This film, along with the work by many others including faculty, staff and students at Maryland, clearly identifies the challenges we still face and the possible solutions to overcome those challenges.”

Hal Daumé III, a professor of computer science with joint appointments in the University of Maryland Institute for Advanced Computer Studies and the Language Science Center, has long been active in research involving bias and fairness in machine learning and AI and will lead the panel discussion.

It will feature Mozilla Foundation fellow Deb Raji, an expert in AI bias whose work is highlighted in the documentary; Margrét Vilborg Bjarnadóttir, an associate professor of management science and statistics in the Robert H. Smith School of Business; Nicol Turner Lee, a UMD sociology lecturer and director of the Center for Technology Innovation; and Adam Wenchel, co-founder and CEO of Arthur AI. 

“Algorithmic bias is something that influences many aspects of our lives,” Daumé said, adding that much of the groundbreaking research in this area has been done by people who come from groups that are often marginalized in computing.

“Broadening their participation—both in discussing these issues and in coming up with answers—is much needed if we truly want to tackle these hard socio-technical problems,” he said.