CONSTITUTIONAL ACCOUNTABILITY OF THE GOVERNMENT FOR MACHINE LEARNING-BASED SYSTEM ERRORS IN DIGITAL PUBLIC SERVICES

Abstract

This study examines the constitutional accountability of the government for machine learning-based system errors in Indonesia’s digital public services. The objective is to analyze how state responsibility should be formulated when digital systems misread data, reject applications, delay access, produce inaccurate classifications, or incorrectly process citizens’ rights. This research applies a qualitative legal method with normative-juridical, conceptual, and socio-legal approaches. The analysis is based on constitutional norms, public service law, government administration law, personal data protection law, electronic-based government regulations, and recent scholarly debates on automated decision-making and public-sector AI governance. The findings show that machine learning-based errors cannot be treated as ordinary technical failures when they affect citizens’ access to public services. Such errors must be understood as failures of public authority because the system operates within the institutional responsibility of the state. Indonesia already has legal foundations for public service, administrative responsibility, digital government, and personal data protection, but it lacks a specific accountability framework for machine learning-based public service errors. This study proposes the concept of state constitutional responsibility for governmental technology failure, consisting of preventive, explanatory, corrective, institutional, and remedial accountability. The contribution of this study lies in framing machine learning errors in public services as constitutional accountability issues, not merely as technical, administrative, or contractual problems.