Generative AI Models as Wicked Resources: A Dynamic Perspective on Resource Governance

The proliferation of generative AI models fundamentally alters organizational capabilities, enabling novel value creation while challenging incumbent governance frameworks. Employing a phenomenon-driven approach, this study integrates and extends property rights theory (PRT) and stakeholder resource-based theory (SRBT) to address governance challenges posed by generative AI. By leveraging system dynamics modeling, we conceptualize the dynamic interplay among stakeholder claims, institutional arrangements, and value appropriation outcomes, highlighting how feedback loops, delays, and accumulations shape these interactions. Our analysis reveals two insights: First, stable equilibrium states in stakeholder claims and property rights arrangements may not invariably lead to equitable outcomes, due to stakeholder power disparities and attribution ambiguity associated with generative AI. Second, framing the evolution of generative AI models as organizational resources from the complementary perspectives of PRT and SRBT reveals distinct resource features largely unexamined in the strategy literature. Hence, we introduce the concept of “wicked resources,” characterizing generative AI models by their inherent attribution ambiguity and emergent unpredictability. Building on prior research on resource complexity and uncertainty in the strategy literature, wicked resources are marked by the difficulty firms face in delineating and enforcing control within shifting sociopolitical contexts. This paper makes three key contributions: addressing the dynamic, multi-stakeholder nature of generative AI governance; introducing wicked resources as a novel resource category in strategy and management literatures; and identifying theoretical gaps, advocating for a dynamic, systemic approach to property rights and stakeholder bargaining.