Who Produces? AI, Delegation, and the Collapse of Managerial Boundaries

Recent advances in generative AI have made a broader AI-mediated reorganization of knowledge work newly visible. In software and adjacent white-collar domains, AI systems increasingly participate in producing artifacts—code, tests, documents, specifications, analyses—while oversight, judgment, and accountability remain assigned to human and organizational actors.This paper argues that the resulting mismatch between who produces, who is recognized as expert, and who remains accountable destabilizes establishedprofessional boundaries, especially between those who execute work and those who coordinate it.The paper’s central claim is that AI redistributes production without equivalently redistributing accountable agency. In human organizations, execution may be delegated while accountability remains shared across recognized persons and roles. With AI systems, by contrast, execution can be displaced into nonhuman agents without creating another socially recognized bearer of judgment, answerability, or obligation. The result is a widening gap between distributed production and distributed accountable agency. This also produces an occupational rupture for workers whose expertise was tied to direct craft.Drawing on work on technological delegation, infrastructure, boundary objects, and professional jurisdiction, and using a small set of public empirical materials from software work, the paper shows how organizations respond to this instability. It argues that AI does not eliminate people management, but destabilizes the older bundling of people management, coordination, andproduction oversight within managerial roles. Organizations respond by building new sociotechnical arrangements—prompts, specifications, review routines,evaluation artifacts, and governance procedures—to restore legitimacy, coordination, and meaning. Using software as a leading case, the paper showshow AI reshapes not only productivity, but the politics of recognition, accountability, and control.Keywords: AI-mediated work; generative AI; large language models; delegation; expertise; professions; infrastructure; accountability; software; STS