Artificial Intelligence-Embedded Restructuring of Enterprise Decision Authority: A Re-examination from the Perspective of Management Theory

Abstract

Artificial intelligence (AI) is becoming deeply embedded in organizational operations, especially in digitally intensive and platform-based business environments. Rather than serving only as a technical support tool, AI is increasingly involved in how decisions are formulated, implemented, and monitored. This development raises an important question for management research: when algorithmic systems gain clear advantages in prediction, optimization, and automated execution, do firms still allocate decision authority in the same way as traditional theories assume ?

Conventional management theories generally treat decision authority as something held by managers and formal governance bodies. These theories are built on assumptions of human-centered decision-making, bounded rationality, and a close connection between authority and responsibility. However, as algorithmic systems begin to shape operational choices more directly, these assumptions become less stable. In many cases, algorithms may not hold formal authority, yet they can exercise substantial influence over outcomes by structuring information, narrowing options, and triggering action.

To explain this shift, this paper develops the Algorithm-Embedded Decision Authority Structure (AEDAS) framework. The framework defines decision authority as a combination of formal authority and effective authority, and argues that authority evolves through three stages: decision support, human-AI co-decision, and layered authority. At the layered stage, algorithms tend to dominate operational decisions, while strategic decisions remain primarily human-led but increasingly constrained by algorithmic systems under a hybrid rationality model. The paper also identifies three mechanisms that drive this restructuring—information power shift, execution automation, and accountability reconfiguration—and develops six propositions together with relevant boundary conditions. Overall, the study argues that AI does not simply improve decision efficiency; it changes how decision authority is distributed inside firms, with important implications for management theory and governance design.