AI-Generated Misinformation in the Electoral Process: A Scoping Review

• In Event: Interventions and Innovations That Impact Turnout and Voting

Abstract

Artificial intelligence (AI) has the potential to dramatically change political campaigning and elections, with significant consequences for democratic participation among voters in countries across the world. In particular, AI may play an increasingly significant role in shaping voter decision-making, especially through its influence on the creation and dissemination of misinformation, which is a crucial and under-explored area of inquiry. In this study, we review the extant literature, including grey literature such as industry reports, to explore the current body of knowledge in this area and identify key gaps for future research. We focus specifically on the use of AI to create higher-quality misinformation in greater quantities at a faster and less cost-prohibitive rate, resulting in high volumes of believable misinformation that may make it exceptionally difficult for the average voter to discern fact from fiction. We look at this specifically within the confines of how AI is being used by political campaigns. We also explore the existing research on discernment, asking whether the tools and skills even exist to separate high-quality, AI-generated political misinformation from actual campaign ads or other political information. Finally, we explore the pathways through which AI-generated misinformation may influence voter decision-making, focusing on areas such as the exploitation of cognitive biases, the creation of belief-based filter bubbles, and the use of highly targeted misinformation to appeal to specific subgroups of voters. The results of this study will be used to inform future research exploring the use of AI by political campaigns.

Authors

• Caroline Orr Bueno, University of Maryland

• Rhyner Washburn, University of Maryland