Georgia Institute of Technology, School of Public Policy
Emory University, Department of Political Science
The use of artificial intelligence (AI) in government services is increasing and poses new opportunities and challenges for governments and citizens. While proponents note AI’s potential to improve public services, practitioners and scholars have raised concerns related to bias caused by AI, a lack of transparency in AI systems, and questions about the role of human oversight versus automated decision-making. Our study introduces the concept of AI-government, and provides evidence from a survey experiment to evaluate how these concerns affect citizens’ attitudes towards the use of AI in public services. The results show that citizens are most concerned about the potential for distributional harms related to bias, with substantively large standardized effects ranging from 0.23 to 0.31 in magnitude. Citizens are relatively less concerned about transparency, and least concerned about human oversight of AI systems. These findings hold regardless of citizens' political affiliation and across governance policy domains. The future, we plan to look more deeply into these ethical concerns and others associated with AI, in order to better understand what mechanisms drive citizen attitudes. This includes looking at different policy sectors, understanding citizen assumptions about particular government agencies or use cases, and seeing what demographic characteristics about citizens matter.