Artificial Intelligence Aversion in Public Policy

Abstract

This dissertation advances our understanding of how individuals relate to the utilization of artificial intelligence within the public policy domain. This is done through theoretical development about the aversion individuals have towards artificial intelligence as well as through empirical examination of that artificial intelligence aversion and how it manifests. In this dissertation I identify that artificial intelligence exists as a unique concept to individuals in contrast to algorithms. I then develop an index to measure people’s levels of aversion to artificial intelligence. I validate that index by assessing how well it does at predicting support for current and future uses of artificial intelligence. Once my index is validated I then turn toward trying to understand what variables are contributing towards people’s different levels of aversion. I first examine the role that perceptions about risk and subjectivity of the area the artificial intelligence is being used in has on people’s aversion index scores. I then examine how demographics influence both perceptions as well as people’s levels of aversion. In these examinations I find that: perceived risk and perceived subjectivity contribute in part to people’s levels of aversion, with perceived risk having a larger effect; and that demographics play a key role in people’s perceptions about the utilization of artificial intelligence. Demographics help to understand how personal levels of aversion to artificial intelligence differ and are identified as an important area of focus if policy makers want to reduce artificial intelligence aversion. This research paves the way for future examination into how aversion changes over time as artificial intelligence is increasingly utilized in people’s lives.