Generative AI chatbots like ChatGPT are increasingly embedded in teens’ everyday routines—not just for academic support, but emotional expression, identity rehearsal, and social interaction. Drawing on interviews with 20 U.S. teens aged 13–18, this paper examines how chatbots are used across academic, emotional, and social domains. While teens often frame these interactions as impersonal or functional, their accounts reveal nuanced forms of self-presentation, tone management, and impression shaping. We introduce the concept of ambient trust to describe how repeated interactions foster a sense of alignment with AI systems—even without deep emotional reliance. We contribute: (1) a thematic account of teens’ cross-domain chatbot practices; (2) a theoretical synthesis drawing on boundary regulation, image management, and related theories of algorithmic mediation; (3) the concept of ambient trust, which helps explain how instrumental use can quietly shape self-expression; and (4) design implications for developmentally appropriate, transparent AI systems.
ACM CHI Conference on Human Factors in Computing Systems