Human-AI Collaboration and the Renegotiation of Managerial Authority: An Integrative Critical Review with a Multi-Case Analytical Framework for Agentic AI Adoption in Industry
TL;DR
This comprehensive review synthesises research on agentic and generative AI, human-AI cooperation, human-AI cooperation, algorithmic management, trust and explainability, decision-rights theory, and technology adoption into a conceptual multi-case framework comparing agentic AI adoption in sectors such as healthcare, finance, manufacturing, retail, logistics, professional services, and IT.
Abstract
The spread of agentic Artificial Intelligence (AI) systems that can autonomously plan, use tools, and perform multi-step tasks instead of offering single recommendations challenges long-held views on where managerial authority lies within organisations. Using a PRISMA-guided search across databases like Scopus, Web of Science, ABI/INFORM, Google Scholar, and industry-specific repositories, this comprehensive review synthesises research on agentic and generative AI, human-AI cooperation, algorithmic management, trust and explainability, decision-rights theory, and technology adoption. It explores how increasing AI autonomy shifts decision-making, accountability, and control from human managers to computational agents. Rather than merely summarising, the review critically organises findings around a five-stage transition from AI as a tool to an assistant, collaborator, decision agent, and finally an organisational actor, highlighting areas of consensus, debate, and gaps, especially in accountability, managerial discretion, and industry differences in autonomy governance. Drawing on theories such as agency, sociotechnical systems, and adoption models (TOE, TAM, UTAUT), the paper introduces a conceptual multi-case framework comparing agentic AI adoption in sectors such as healthcare, finance, manufacturing, retail, logistics, professional services, and IT. It proposes a model linking key factors influencing authority shifts and posits ten propositions for future testing. The review provides a shared vocabulary for authority shifts in AI contexts, extending control theories into AI environments, and guiding managers, boards, and policymakers in developing delegation, escalation, and accountability strategies. While the framework is conceptual and propositional without empirical validation, it outlines a future research agenda aimed at understanding how organisations can maintain meaningful human oversight as AI agents play a larger decision-making role.