Epistemic Capabilities of Artificial Intelligence for Public Administration and Policy Research: A Framework for Scholarship in the Age of Intelligent Machines

ABSTRACT Artificial intelligence (AI) has gained much interest in public administration and public policy fields in recent years. Research on and that uses AI has so far treated it as an object or context within the broader administrative and governance discourse. In this article, we aim to shift the locus of scholarly attention on AI by focusing on its premise as a new epistemic technology that will shape how scholars conduct research. In doing so, it proposes a framework for leveraging AI as a methodological enabler that augments researchers' cognitive and analytical capabilities, thereby overcoming longstanding methodological barriers in the fields. We propose a framework distinguishing AI applications along two epistemological dimensions of research that make use of synthetic data and real‐world data: (1) inductive approaches to discover patterns and build theory to address exploratory questions (“what,” “so what”), and (2) deductive approaches to test hypotheses and establish causal relationships to serve explanatory and predictive purposes (“why,” “how”). We contend that AI has propelled the fields toward a “new normal science” characterized by human‐machine collaboration.