US senators ask online proctor firms for evidence they are fighting biometrics bias

Allegations of bias in proctoring software will never be the motivating controversy that is bias in policing facial recognition software, but it will be held up as another example of the technology’s Achilles’ heel.

Last week, six Democratic senators tried to buttonhole executives of three online examination firms on the topic. The politicians want the heads of ExamSoft, Proctorio and ProctorU to tell them how each protects students’ civil rights and how their apps work with everyone equally.

The subtext is that facial recognition systems too often misidentify people of color and those wearing religious head coverings. Proctoring software authenticates test takers and watches for evidence of cheating. People with certain physical, mental and emotional conditions can, indeed, act in ways that can make even a human proctor suspicious.

Some students see the potential problem more broadly. They do not want to be digitally surveilled to begin with.

Opponents of the growing practice say they are being presumed guilty by being required to use the app. And as more classes go online, they object to the fact that anyone or anything in the background, even briefly, is recorded. All the information is then stored for a period during which it can leak.

Someone, in fact, hacked into Verificient Technologies’ Proctortrack in October. Company executives said no personally identifiable information was accessed.

The senators requesting information are Cory Booker of New Jersey, Elisabeth Warren of Massachusetts, Richard Blumenthal of Connecticut, Ron Wyden of Oregon, Tina Smith of Minnesota and Chris Van Hollen of Maryland.

In letters to the vendors, the senators say tic disorder or muscle reflexes have resulting in some students being flagged following a test. The company have until December 17 to reply.

Article Topics

accuracy  |  AI  |  algorithms  |  biometric bias  |  biometric identification  |  biometric monitoring  |  biometrics  |  facial recognition  |  remote proctoring  |  United States