The paper discusses the fundamental problems related to determining the subject of administrative competence and establishing the structure of the competence norm in cases where the legislator has fully or partially delegated the process of applying the law to artificial intelligence systems. The main objective of the analysis is to consider the possibility of changing the understanding of administrative competence and to assess the constitutional and administrative aspects of the legal and democratic legitimacy of delegating competences to apply the law to artificial intelligence systems. Separate considerations are devoted to the issue of the limits of the admissibility of modifying or dispersing the legal responsibility of the State for the use of specific algorithmic systems in the processes of applying administrative law. The concluding remarks conclude that ensuring that the operation of algorithmic decision-making systems in administrative processes of law application complies with the principles of democratic legitimacy, formal rule of law and legal accountability of public administration requires maintaining the competence of legislative authorities to decide on the principles and procedure for introducing artificial intelligence into the sphere of concretisation or implementation of administrative law norms and to establish very detailed rules for the design, construction, implementation, validation, control and modification of the essential elements of such systems.