Citizens' perceptions of accountability in AI-based public sector decision making

Research has increasingly focused on identifying and analyzing the key components of accountability in the use of artificial intelligence (AI) in the public sector. However, much of the existing literature takes a government-centric approach, emphasizing how public organizations implement accountability measures while paying limited attention to citizens' perspectives. This study addresses this gap by empirically examining whom citizens believe should be held accountable and what information about AI-based decisions should be communicated to the public. Using data from a nationally representative sample of U.S. citizens, our analysis reveals several overall trends in public perceptions of accountability for AI-based systems, as well as demographic differences. These findings provide valuable empirical insights to advance research and practice on the use of AI in government.