Algorithmic Determinism vs. Executive Agency: A Theoretical Examination of Strategic Management in the Age of Artificial Intelligence

Abstract

Strategic management settled its dispute between environmental determinism and managerial choice by treating constraint as arriving from outside the firm. Artificial intelligence unsettles that arrangement, because the algorithms that filter markets, rank acquisition targets and evaluate strategic alternatives are installed by executives themselves. This conceptual paper asks where executive agency goes when strategic decisions pass through algorithmic systems. Drawing on upper echelons and managerial discretion research, sociomaterial accounts of technology and evidence on how decision-makers respond to algorithmic outputs, it develops a framework of configured discretion. Algorithmic determinism is theorized as an endogenous, cumulative constraint operating through attention, evaluation and capability. Executive agency is relocated into three practices: framing objectives, contesting outputs and reconfiguring decision architecture. The central argument is that delegation is self-binding. Each act of delegation can erode the capabilities executives need to contest the system later, so formal authority persists while real latitude shrinks. Seven propositions follow, with a state-transition model of four adaptation states (algorithmic capture, governed hybridity, executive voluntarism and fragmented drift) that recasts the two-continua model of organizational adaptation for firms whose constraints are partly self-made. The framework reconciles conflicting findings on algorithm aversion and appreciation and on the value of human override. Governance regimes moderate the loop: fiduciary duties of care and oversight, audit-committee benchmarks against core models and party-state oversight each shape how far effective discretion can fall below formal authority.