WE NEED TO TALK: CONVERSATIONS ABOUT MENTAL HEALTH, SELF-DISCLOSURE, AND THE ROLE OF AI CHATBOTS

Abstract

This dissertation examines how college students engage in conversations about mental health, focusing on the benefits and challenges of self-disclosing to different conversational partners. Recent research highlights the potential of AI-powered chatbots to improve mental health through greater accessibility (Alanzi et al., 2024; Boucher et al., 2021; Denecke et al., 2021; Kim et al., 2024; Li et al., 2024; Wester et al., 2025). Drawing on the disclosure processing framework, which suggests that non-human agents may reduce perceived judgment and impression management concerns (Ho et al., 2018), as well as foundational theories of impression management (Brown & Levinson, 1987; Goffman, 1959), this dissertation develops and empirically tests an integrated model of disclosure processing.This dissertation includes three studies. First, an interview explores existing beliefs and past experiences surrounding conversations about mental health across AI chatbot and peer contexts. Second, a survey examines associations between mental health beliefs, well-being indicators, and individual characteristics and willingness to disclose, intention to initiate conversations about mental health, and conversation behavior. Lastly, an experiment investigates the effects of mental health disclosure through two distinct conversational partners: anonymous peers in computer-mediated environments and AI chatbots. This dissertation identified salient beliefs for AI chatbot as providing reassurance, knowledgeability, and solution provision, along with descriptive norms about friends, as positively associated with mental health communication outcomes. For peers, beliefs emphasizing relatability and support, along with descriptive norms about friends, were most strongly associated in the positive direction with these outcomes. Experimental results further show that, whereas peer interactions were shaped by barriers such as public stigma, self-stigma, and social well-being, these constraints were not observed in AI chatbot interactions, underscoring their potential as less stigma-constrained contexts for mental health conversations. This study advances understanding of college mental health by examining how conversations about mental health occur and the perceptions behind what facilitates or hinders such conversations. Additionally, it offers insights into the role of AI in mental health discourse, providing implications for digital mental health interventions. By identifying effective methods to encourage mental health discussions and understanding the mechanisms behind disclosure to AI chatbots, this research informs strategies for encouraging an open communication environment within campuses.