Why hesitant help-seekers consider AI: correlations of mental health chatbot adoption among Chinese university students

Purpose Despite the well-documented reluctance of individuals, particularly university students, to seek professional psychological help, this study aims to seek out the potential of AI to bridge service gaps for those hesitant help-seekers. This study adopted an exploratory approach using a broad student sample to investigate psychosocial and technological factors influencing adoption attitudes toward AI, revealing unique adoption patterns that may differ from conventional mental health supports. Design/methodology/approach The study employs a quantitative approach, utilizing a survey methodology. Data were collected from 323 university students and analyzed using multiple regression and mediation analysis (SPSS) to identify predictors of willingness to adopt ChatGPT for mental health support. Findings The results reveal that 68.7% of participants were open to using ChatGPT for mental health purposes, with perceived effectiveness being the strongest predictor (ß = 0.70, p < 0.001). Interpersonal relationships and self-reported mental health status were indirectly associated with adoption intentions through perceived effectiveness. These exploratory results suggest that AI tools may engage traditionally hesitant populations differently than conventional services, though the observed relationships require verification through controlled studies. Originality/value As one of the first exploratory studies examining AI chatbot adoption for mental health support among Chinese university students, this research provides foundational insights into a nascent field. The findings illuminate key psychosocial mechanisms (e.g. the strong positive association of perceived effectiveness) and technological considerations that may distinguish AI-based support from traditional help-seeking paradigms. While revealing technology-specific adoption patterns, this study claims these correlational findings require experimental validation.