Beyond Digital Companionship: Rethinking LLM Chatbots and Adolescent Loneliness Through a Youth-Centered Lens

Abstract

Teenage loneliness is a prevalent issue in digitally saturated societies, where pervasive online connectivity often fails to resolve emotional disconnection. This ongoing youth-centered study critically examines how LLM-powered chatbots shape loneliness among adolescents. Grounded in a migrant adolescent's lived experience and supervised academic development, the research prioritizes youth perspectives alongside theories from social computing, developmental psychology and human–computer interaction. It analyses chatbot effects across four adolescent groups: general youth, individuals with anxiety or depression, neurodivergent adolescents, and immigrant youth. Chatbots deliver short-term benefits by enabling low-pressure emotional expression, language practice and social skill rehearsal, yet they may aggravate long-term loneliness via psychological dependency, reduced human social engagement and inadequate crisis detection. The study also highlights structural risks, including engagement-oriented algorithmic design, subtle emotional manipulation and unequal access to supportive functions. Three targeted design implications are proposed: population-adaptive personalization, graduated AI-to-human support escalation pathways, and transparent disclosure of AI limitations. Future research will expand this participatory model to diverse adolescent samples to empirically validate the proposed framework.