АЛГОРИТМДІК ӘКІМШІЛІК ШЕШІМДЕРДІҢ УӘЖДІЛІГІ, ТҮСІНІКТІЛІГІ ЖӘНЕ ОЛАРДЫ СОТ ТӘРТІБІМЕН ДАУЛАУ МӘСЕЛЕЛЕРІ

Abstract

The article examines the administrative-law problem of algorithmic administrative decision-making: whether an act of public administration can be regarded as lawful, reasoned and judicially reviewable where its factual or evaluative basis is produced by an algorithmic system, while the affected person receives only the final outcome. The study is based on the Administrative Procedural-Processual Code of the Republic of Kazakhstan, the Law of the Republic of Kazakhstan “On Artificial Intelligence”, legislation on informatization and personal data, Kazakhstan’s 2024–2029 Concept for Artificial Intelligence Development, international instruments and the SCHUFA, SyRI and State v. Loomis cases. The methodology combines formal legal analysis, comparative legal analysis, systemic interpretation, doctrinal analysis and legal modelling. The article argues that technical explainability and legal reason-giving are not identical. Explainability describes the operation of a model; reason-giving enables the addressee to understand, verify and challenge the exercise of public authority. The author proposes a model of a “minimum judicial file” for algorithmic administrative decisions, including the legal basis for using the system, the role of the algorithm, relevant data, criteria, logs, human oversight and the responsible public official.