Parasocial interaction with artificial intelligence (AI companions): association with social anxiety and depressive symptoms in youth

Abstract

The purpose of the study was to determine the relationship between the level of parasocial interaction with AI companions, social anxiety, and symptoms of depression, anxiety, and stress among young people, as well as to discuss the potential compensatory role of such interaction in the context of psychological distress. Materials and Methods. An empirical study was conducted in 2026 in Ukraine using an anonymous online survey via Google Forms. The sample comprised 127 respondents aged 17 to 35. The psychodiagnostic toolkit included: the Adapted Parasocial Interaction with AI Companions Scale (PSI-AI); the Liebowitz Social Anxiety Scale (LSAS-SR) to assess social fear and avoidance; and the Depression, Anxiety, and Stress Scale (DASS-21). Results. Empirical data revealed that 89.8% of respondents had successfully integrated AI agents into their micro-social environment. This indicates a fundamental shift from the classical one-way parasocial model (typical for mass media or television) toward a novel paradigm of two-way, interactive, and personalized dialogue featuring the characteristics of a natural emotional contact. High concentrations of destructive psycho-emotional states, driven by the prolonged wartime state in the country, were recorded within the sample. Specifically, elevated levels of general anxiety were detected in 68.5% of respondents (with 33.1% exhibiting severe and extremely severe clinical forms). Signs of pronounced social anxiety were found in 64.6% of individuals, while depression levels above the norm were observed in 58.3%, indicating a significant depletion of the adaptive capacity of youth’s psyche. Concurrently, a moderate-normative level of chronic stress (50.4%) demonstrates a habituation effect – a partial psychophysiological adaptation to conditions of continuous threat. A profound data analysis using the PSI-AI scale explicated an important psychological pattern: existing anxiety symptoms block real interpersonal communication due to the fear of negative evaluation, causing the need for communication to be channeled into a safe digital space. In this environment, the AI companion serves as an effective tool for daily psycho-emotional self-regulation and compensatory security (M = 8.70). AI is perceived as a non-judgmental, empathetic interlocutor trusted with the most intimate personal experiences (M = 8.25). However, low emotional attachment scores (M = 7.26) clearly illustrate that, at the current stage, youth lack a pronounced psychological dependence or total substitution of real interpersonal connections. Conclusions. The empirical study provided a systemic description of the specifics of modern youth’s interaction with generative AI technologies through the prism of their mental health. It is proven that interaction with digital companions currently functions as an additional, adaptive coping resource during periods of acute psycho-emotional tension, rather than a full-fledged substitute for real relationships. Nevertheless, the identified stable positive correlations between the level of involvement in communication with AI and the severity of anxiety-depressive symptoms carry a hidden risk of transforming a quick self-help tool into a destructive mechanism of escapism and reality avoidance. Based on the results, practical recommendations were formulated for mental health professionals on the necessity of clearly differentiating between the adaptive use of AI as an emotional buffer and destructive social withdrawal. The study substantiates the feasibility of integrating specialized digital mental hygiene blocks into higher education curricula and national psychosocial support programs to foster youth’s mindful perception of AI solely as a temporary communicative simulator rather than a full replacement for interpersonal relationships.