Multi-Domain Self-Esteem Profiles Associated with Adolescent Engagement with Conversational AI: A Latent Profile Analysis

This study examines whether multidomain self-esteem profiles differentiate adolescents’ interactions with conversational AI. Data come from a national school-based survey of Czech adolescents aged 13–17 (N = 42,772). Latent profile analysis based on Home, School, and Peer self-esteem identified three groups: high, moderate, and low self-esteem. These profiles were compared across AI-related domains, including social substitution, emotional regulation and self-disclosure, intimacy-related use, and perceived ease and safety of AI communication.Adolescents with low self-esteem reported the highest levels of friend-like interaction with AI, stronger beliefs that AI can replace human relationships, more frequent use for emotional support and disclosure, greater intimacy-related engagement, and stronger perceptions of AI as easier and safer than human communication. They also reported the highest psychological distress.