CAREER: The Economic and Social Consequences of Frontier AI Innovation

Harvard University, Cambridge MA

Investigators

Abstract

This project uses large administrative data sets and innovative data analysis methods to study the social and economic consequences of innovations in artificial intelligence (AI). The research seeks to determine conditions under which societies that respect individual freedom can lead the world in rapid, efficient, and socially beneficial AI innovation. A second part of the project investigates the impact of AI-led innovation on citizen privacy. Finally, the project will consider whether and how global trade in AI-enabled products could be a risk to individual freedom. Specifically, the research studies conditions under which direct government control can coexist with innovation and whether citizens may be able to use AI to protect civil liberties. The project also considers whether global trade in AI technologies may create risks to these liberties. This proposed research will use three projects to study the interaction between technological innovation and centralized government control of economic and civil life. The first project will use a large data set to investigate whether or not AI innovation might be stimulated by a highly centralized regime’s desire for political control. The second project combines field experiments with administrative data to study how citizens change their behavior when a large number of daily actions are explicitly monitored by their government, and whether such experience shapes privacy preferences. The third project will build a large data set on AI exports and use the data to investigate whether: (i) representative governments have a comparative disadvantage in AI technology, and (ii) countries importing AI technology from non-representative governments subsequently increase political suppression. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.