Artificial Intelligence and Democracy's Information Problem

Democracy in America faces a fundamental contradiction that threatens its future. Elected officials do not respond to the wants and needs of most Americans, yet they are being reelected at historically high rates. That contradiction has fueled rising dissatisfaction with democracy and support for its more autocratic alternatives. To date, legal scholars have neglected a critical source of that accountability deficit: democracy’s information problem. Most democratic citizens lack the capacity or competency to acquire and process information about the policies their elected officials support or oppose and how those policy choices impact their lives. Theoretically, people may draw on increased education and use information shortcuts like party labels to ameliorate their lack of policy knowledge. But the United States’ widening education disparities and partisan polarization undermines those strategies. As a result, citizens cannot effectively hold politicians to account for decisions that undermine their well-being. Democracy desperately needs another tool to respond to its information problem, or it may not survive. In this Article drawn from my Constitution Day Lecture at the University of Pittsburgh School of Law, I argue that artificial intelligence (“AI”) can serve as that tool. While many consider AI to be a threat to democracy, it is a technology uniquely capable of making democratic information easy to acquire and process. If properly developed, AI can help democratic citizens hold elected officials accountable for the policy decisions they make. For AI to effectively mitigate democracy’s information problem, though, it must be a trusted source of information. To avoid AI’s potential perils, we should utilize the principles of “trustworthy AI” to build a legal and institutional framework that conserves the democracy-promoting uses of the technology.