A bridge, not a destination: YouTube viewer perspectives on AI mental health support and human therapy

Artificial intelligence (AI) tools are increasingly used for mental health support, yet little is known about how they are understood outside clinical trials and survey-based research.

This study examined public perceptions of AI mental health support through a convergent mixed-methods analysis of 7,949 YouTube comments posted across ten videos discussing AI and mental health. Quantitative analyses included VADER sentiment analysis, NRC emotion profiling, latent Dirichlet allocation topic modelling, and keyword co-occurrence network analysis. Qualitative analysis used Braun and Clarke's reflexive thematic analysis to identify patterns of shared meaning within a purposive sample of high-engagement comments.

Overall sentiment was moderately positive (57.93%), but this positivity was qualified by substantial negative sentiment (24.58%) and recurring emotional signals of both trust (14.98%) and fear (7.92%). Topic modelling showed that the dominant lines of discussion centred on unmet support needs and the question of whether AI should replace human therapists. The thematic analysis generated five themes: AI as a bridge where human care is absent, AI as comforting but overly validating, AI as unable to reproduce authentic human therapeutic encounter, AI as a possible driver of social disconnection, and AI as a site of privacy and commercial concern.

Taken together, the findings suggest that public responses to AI mental health tools are best understood as conditional rather than absolute. Commenters valued AI when it provided immediacy, low-cost access, and a low-risk space for disclosure, but resisted it where it appeared to threaten empathy, relational depth, or privacy. The study contributes a large-scale, naturalistic account of public reasoning about AI mental health support and highlights implications for digital mental health design, governance, and human-in-the-loop care.