THE PHENOMENON OF ARTIFICIAL INTELLIGENCE ADDICTION (AI ADDICTION): THEORETICAL ANALYSIS, RISK FACTORS, AND RESEARCH PERSPECTIVES

Abstract

Abstract. The rapid and large-scale integration of generative artificial intelligence (AI) systems into daily life creates an unprecedented context for the formation of new forms of behavioral addictions. Unlike previous digital technologies, AI offers unique interaction mechanisms with a potentially much higher addictive potential. This poses a serious threat to mental health, especially among vulnerable populations, and necessitates the development of a new conceptual framework for understanding, diagnosis, and therapeutic intervention. Objective. To conduct a theoretical analysis of the AI addiction phenomenon, identify its unique phenomenological features, propose hypothetical neurobiological and psychosocial mechanisms of its formation, and outline the main diagnostic challenges and future research directions. Materials and Methods. A theoretical review and conceptual analysis were conducted based on a systematic search of scientific literature in the PubMed, Scopus, Web of Science, and Google Scholar databases for the period from 1996 to 2025. A narrative synthesis method was used for data analysis and generalization. Results. The literature analysis identified three key characteristics of AI addiction: 1) hyper-personalization of content, creating a powerful, individualized reward system; 2) anthropomorphism and the simulation of social interaction, allowing compensation for deficits in real social connections; and 3) continuous gamified interaction loops that exploit dopamine reinforcement loops similar to gambling. A neurobiological model is proposed, linking these stimuli to the dysregulation of the mesolimbic dopamine system, hypoactivity of the prefrontal cortex (leading to reduced cognitive control), and alterations in the default mode network. Psychosocial risk factors, including loneliness, social anxiety, low self-esteem, and deficits in interpersonal communication skills, are examined. Diagnostic challenges related to symptomatic overlap with other addictions and the absence of validated assessment tools are outlined. Conclusions. AI addiction is an emerging potential behavioral disorder with a unique pathogenetic profile that distinguishes it from existing digital addictions. Urgent further research, including longitudinal studies and neuroimaging experiments, is required to verify the proposed models, develop validated diagnostic criteria (e.g., for inclusion in future versions of the DSM/ICD), and create evidence-based prevention and therapy programs.