When AI Performs Journalism: Delegation and Journalistic Roles in Collective News Production

Abstract

Generative artificial intelligence (AI) is increasingly involved in selecting, interpreting, and generating news content. This conceptual article examines what that involvement means for journalistic role theory and news production. Building on the journalistic role performance framework, we argue that AI can materially shape role performance without possessing a corresponding role conception or perception. Furthermore, we introduce the concept of the architecture of delegation to explain variation in this relationship: the arrangement by which content-shaping operations are assigned to AI, normative orientations are translated into technical instructions, and outputs are incorporated into news. Role performance should align less closely with normative orientations as AI takes on broader, more discretionary, and more consequential functions, and more closely when those orientations guide its use and outputs are substantively reviewed and traceable. These effects are role-specific because their direction and magnitude depend on how each operation affects the practices through which particular roles become manifest. Depending on these conditions, delegation may amplify or displace roles or alter their combination, and may produce performances that were neither intended by journalists nor meaningfully encoded in organizational instructions. We further argue that, as AI-mediated performances become routinized and organizationally rewarded, they may feed back into professional culture and reshape normative orientations.