Generative AI and the Nationalization of US Politics

Abstract Artificial Intelligence (AI) was seemingly everywhere by the end of 2024, and the 2024 US presidential election was the first American national election to be conducted wholly in an AI era. Nevertheless, relatively little is known about how effectively generative AI contributes to learning about politics. This study explores that question in the context of research on subnational US politics. Based on a novel methodology that combines the analysis of AI-generated profiles on several US states with interviews with state-level experts, this article identifies and analyses a prevalent national bias in the state-level content produced by generative AI. This bias is both a consequence of and a contributor to the problem of the nationalization of American politics, which itself undermines the principles of federalism that undergird Madisonian democracy in the United States.