Education and science group: Proctoring industry should straighten up
A prominent computing society has issued guidelines — although not standards — for the remote-test industry, including in schools.
The Association for Computing Machinery has published a lengthy list of recommendations on equity, privacy, security, accessibility and efficacy for exam proctoring systems that feature facial recognition and other biometric capabilities.
These invigilation tools had a moment in the sun with the advent of COVID when remote testing became the rule rather than the exception. For the most part, they administer tests and monitor university test takers.
But they came in for criticism over foreseeable biometric shortcomings including the deployment of products that are confused by the faces of people who are not white males.
The association, with 100,000 members globally, claims to be the largest such educational and scientific group. It is in a good position to both help regulate remote testing and to work with other associations faced with similar puzzles, namely automating identity verification and fairly monitoring for anomalous events in video feeds.
Beyond boilerplate guidelines that every organization should already be following (destroy sensitive data when no longer needed, tell test takers how data will be collected and managed, etc.), association members identified the need to separate two critical data feeds.
A wall is required between biometric data and test responses, according to the group, to assure the privacy of test takers.
And ability to pay service fees should not prohibit test takers from participating. The association says schools, for example, sometimes waive lab fees for students who are less financially secure, and the same accommodation should be made in this instance.
In a pointed jab, the association say vendors need to do more for end-to-end security and be open about known problems and risks. In fact, organizations should wrestle “affirmative statements that vendors will not suppress warnings about defects.”
Nor should vendors use legal strategies to hide legitimate product criticism.
ACM | biometric data | biometrics | data collection | data protection | facial recognition | monitoring | privacy | regulation | remote proctoring