Session: Generative AI In Mental Health Pathways and Services: Social Connectedness, Self-Disclosure, Help-Seeking, and Well-Being (Society for Social Work and Research 31st Annual Conference)

Generative AI is rapidly entering the mental health landscape across adolescence and emerging adulthood, yet its role remains poorly understood. Existing discussions often emphasize either promise or risk without clarifying how generative AI relates to social connectedness, emotional disclosure, formal help-seeking, and mental health outcomes across developmental stages. This symposium brings together four papers that examine generative AI in mental health pathways from complementary angles, moving from evidence synthesis to adolescent mechanisms to college student help-seeking and well-being. Together, the papers address whether generative AI functions as a bridge to human support and formal care, a parallel coping resource, or a potential substitute that may alter help-seeking trajectories.

The first paper presents a systematic review and meta-analysis on large language model use and human-human social connectedness. Because social connectedness is upstream of help-seeking, this paper evaluates whether AI interaction appears to support or undermine interpersonal processes relevant to care engagement. The findings point to an important evidence gap: although structured interventions may improve selected interpersonal outcomes under controlled conditions, there is still limited evidence on how everyday AI use relates to human connection and movement toward care.

The second paper focuses on adolescents and examines whether self-compassion operates differently for youth with and without ADHD in shaping disclosure to AI. This study highlights that AI-based mental health engagement is not neurotype-neutral. It shows that mechanisms supporting disclosure may differ across groups and that assumptions about universal digital mental health resources may overlook the needs of adolescents who require more structured supports.

The third and fourth papers focus on U.S. college students using the same Healthy Minds Study AI module but address distinct questions. One paper applies Andersen’s Behavioral Model of Health Services Use to compare predictors of traditional therapy use and generative AI use for mental health. Findings suggest that therapy use is more strongly need-driven, whereas AI-based mental health use is shaped more strongly by enabling beliefs, especially trust in AI. The other paper uses latent class analysis to identify multi-domain profiles of generative AI use and examines associations with depression, anxiety, and loneliness. Results indicate that broader and heavier patterns of AI engagement are linked to greater mental health burden, whereas lighter, more bounded patterns of use are associated with comparatively more favorable outcomes.

Taken together, these papers offer a cohesive account of how generative AI may operate before, alongside, or apart from formal care. The symposium integrates social connectedness, disclosure, help-seeking, and mental health outcomes across two developmental periods and multiple methodological approaches. In doing so, it advances a social work agenda for understanding generative AI not simply as a technology, but as an emerging part of mental health pathways that may shape how young people connect, cope, and seek support. The symposium is relevant to social work researchers, educators, and practitioners seeking evidence to guide ethically grounded, developmentally informed, and service-aware responses to generative AI in mental health contexts.