Abstract
ABSTRACT Generative artificial intelligence (AI) now performs many explicit tasks in organizational knowledge creation, such as drafting, summarizing, and combining documents. This paper asks what this change means for the human agency on which knowledge creation depends. The paper starts from the position that AI does not possess tacit knowledge. Its concern is that AI can automate explicit tasks through which people have traditionally developed tacit knowing. The study uses a conceptual theory‐building design supported by a structured literature review. A Scopus search returned 192 records, and two inclusion criteria were applied. Together with three contributions found through citation tracking, a final set of 15 models of organizational knowledge creation with a defined role for AI was analyzed. The review shows that, despite assigning AI different functional roles, these models continue to depend on human actors for embodied tacit knowing. However, they rarely examine how AI use may change the way people form tacit knowing over time. Building on this gap and on empirical studies of automation and learning, the paper introduces the Human Agency Deficit. It is defined as the gap between the human agency an organization needs to evaluate, own, and extend its knowledge and the human agency that its AI‐supported work practices actually develop and use. The deficit is treated as a conditional risk, not an inevitable outcome. It is mapped across the SECI spiral as a reinforcing loop, and boundary conditions are specified under which AI works as a learning scaffold. The paper then proposes a three‐dimensional process framework with observable indicators, an illustrative application, and five testable propositions.