Governing the Algorithmic State: Artificial Intelligence, Public Administration and the Reconfiguration of State Capacity in Contemporary Africa

Abstract

Artificial Intelligence (AI) is increasingly embedded in public administrations as African governments pursue digital transformation, reform, and improved public services. This expansion raises a theoretical problem because existing approaches often treat AI as a technical instrument for efficiency, prediction, automation, or decision support. These approaches offer limited explanation of how AI becomes institutionally embedded and reshapes administrative capacity. This article develops Algorithmic Administrative Embeddedness (AAE) as a conceptual construct for explaining this process. The analysis adopts a theory-building approach based on a synthesis of public administration, institutional, sociotechnical systems, and public value perspectives. AAE comprises five dimensions: algorithmic decision support, administrative capacity augmentation, institutional accountability, public value creation, and citizen inclusion. These dimensions identify how embedded AI redistributes knowledge, discretion, authority, and accountability. The framework proposes that AI reconfigures state capacity through interactions among officials, institutions, data, and algorithms. It develops theoretical propositions linking institutional governance conditions to consequences of algorithmic embedding. The framework explains why AI can strengthen capability under robust governance while reproducing inequality, obscuring responsibility, or reinforcing institutional weakness under fragile governance. The article contributes an Africa-centred conceptual foundation for analysing algorithmic state capacity and propositions to guide future empirical research on algorithmic administration across public institutions.