Algorithmic Legitimacy and the Digital State: Rethinking Governance, Trust, and Accountability in the Age of AI

Abstract

This conceptual article examines how public authority is legitimised when administrative decisions are shaped by algorithms, automated decision systems, and data infrastructures. It develops an analytical framework that translates classical accounts of legitimacy into the institutional conditions of algorithmic governance. Easton’s systems analysis provides the feedback logic; Scharpf’s input/output distinction and Schmidt’s throughput legitimacy clarify participation, process quality, and performance; Luhmann, Lipset, Wesołowski, and Biernat illuminate self-reference, belief, reflective action, and constitutive legality. The article defines algorithmic legitimacy as the publicly defensible authority to govern through computational systems. Such authority depends on representative problem framing and data practices, documented and contestable model design, auditable outcomes, and effective routes to explanation, appeal, correction, and redress. It argues that technical accuracy or administrative efficiency cannot by themselves establish legitimacy, particularly in high-stakes public decisions. Instead, legitimacy must be renewed across the data lifecycle through public reason-giving, independent scrutiny, and institutional capacity to correct error. The article’s contribution is a multidimensional, testable framework in which algorithmic legitimacy is reflexive by design rather than presumed from technical performance.