Key points are not available for this paper at this time.
The relevance of this study stems from the rapid introduction of artificial intelligence (AI) technologies into criminal proceedings amid the lack of a unified conceptual framework. This terminological uncertainty creates systemic legal risks, ranging from the uncritical acceptance of algorithmic results as absolutely reliable to the difficulties in challenging decisions made using AI. Thus, the development of doctrinal definitions is no longer a purely theoretical task and is becoming a vital practical requirement for the safe integration of AI technologies into criminal proceedings. The subject of this study is the analysis and structuring of existing approaches to defining artificial intelligence, as well as the delineation of key concepts such as "artificial intelligence," "AI system," and "AI assistant/agent" for the purposes of their subsequent effective legal regulation in criminal proceedings. This study utilizes terminological, comparative legal, and critical analysis of existing scientific and legislative definitions of AI. The scientific novelty of this study lies in its proposal of an original three-level hierarchy of concepts, specifically adapted to the needs of criminal proceedings. The author consistently distinguishes: 1) AI as a field of knowledge and technology; 2) an AI system as a specific software product – the primary object of legal regulation; 3) functional roles (AI assistant and autonomous AI agent), classified according to the degree of autonomy in performing procedural actions. The main conclusion is that legal regulation should shift from the abstract concept of "artificial intelligence" to specific AI systems, establishing a differentiated legal regime depending on their functional role and level of independence. The degree of autonomy of the system is proposed to be the key regulatory criterion. Consolidation of this system of definitions will create the necessary terminological foundation for a balanced legal regime that will realize the potential of AI to improve the efficiency of legal proceedings without violating fundamental procedural guarantees.
Aram Edvardovich Arzumanyan (Thu,) studied this question.