We propose a harm-centric conceptualization of privacy and op- erationalize it in the context of using artificial intelligence (AI) in education and employment. In an online study (N=400), US college and university students reported their perceptions of 14 harms (e.g., manipulation) when AI infers personal data (e.g., demographics and personality traits) and use it in decision-making. We demonstrate that our approach can reliably and consistently measure privacy, sidesteps many limitations in existing frameworks, and captures harms from modern technology that would remain undetected by other frameworks. We surface nuanced perceptions of harms, both across the contexts and participants’ demographic factors. Based on these results, we discuss how privacy can be im- proved equitably and inclusively. This research extends privacy theory and provides practical guidance to improve privacy in vari- ous technology use domains.
ACM CHI Conference on Human Factors in Computing Systems