Critical Questions for Facial Recognition Technologies in Higher Education: A Handbook for Stakeholders and Decision-Makers
Authors
Jillian Kwong
Stefanie Demetriades
Ali Rachel Pearl
Noy Thrupkaew
Jef Pearlman
Colin Maclay
Executive Summary
Adoption of facial recognition (FR) technologies has rapidly expanded across multiple sectors of society in recent years, including in higher education. These technologies, which employ various forms of machine learning and artificial intelligence to detect, identify, or analyze individual faces, enable an unprecedented scope of surveillance. These capabilities are often presented by tech companies selling FR tools as powerful automated solutions for campus security and student monitoring. In the sudden disruption to campus life caused by the COVID-19 pandemic, for instance, many universities have expanded the reach of these technologies as they turned to FR-powered systems like Respondus and Proctorio to fill gaps in identity authentication and exam proctoring via facial detection. Using automated facial recognition systems for traditional proctoring or security purposes may seem like familiar territory to university administrators, but it is not.
The use of FR technology introduces new, non-obvious dangers and potential sources of discrimination, and must be evaluated alongside all other uses of FR. As these systems continue to rapidly expand into unanticipated applications—particularly in times of crisis and uncertainty—institutions are often ill-prepared to fully evaluate the potential implications and substantial risks of FR adoption. In particular, uncritical adoption of these technologies could directly harm members of the university community, while normalizing dangerous new forms of surveillance, racial profiling, and automated monitoring of people in public.
One might be inclined to argue that FR systems are always “learning” and therefore “improving,” but these same systems are watching and profiling students, as well as enrolling students in testing their companies’ products, without student consent or compensation.
We intend for this document to serve as a guide to the issues surrounding FR—which includes our analysis and subsequent rationale for banning FR—for university administrators and stakeholders as they work to make informed decisions about facial recognition in their classrooms and communities. We know that universities will have to return over and over again to decisions around whether or not to adopt new technologies, and we hope that this guide will serve as a blueprint and provide administrators with tools for evaluating the hidden risks of adopting future, as-yet undeveloped technologies.
Drawing on a wide range of scholarship, journalism, technology, and policy sources, we identify the following key categories requiring careful consideration: a) efficacy, b) security risks and hidden costs, c) data protections and consent, d) racial and social justice, and e) institutional capacity.
In each of these categories we find either insufficient justification to warrant FR adoption, or, more concerningly, evidence that FR poses serious danger to the interests of the campus community and environment. On the basis of our evaluation, we recommend in the strongest terms that universities do not adopt facial recognition and terminate existing uses of the technology in and around the campus community.
This publication grew out of the MASTS "Mobilizing Controversy" working group.