Benefits over risks: a technology acceptance study of LLM-powered conversational agents for non-clinical mental health support

Large language models (LLMs) have enabled conversational agents for mental health-related applications outside clinical settings, yet user acceptance remains insufficiently understood. Through survey data from 202 participants, we examine how perceived benefits, risks, trust, and anthropomorphism influence initial acceptance intentions towards LLM-powered conversational agents for non-clinical mental health support. Hierarchical regression analysis showed that perceived benefits were the primary drivers of acceptance, with performance expectation and hedonic motivation emerging as the strongest predictors. Female participants reported higher acceptance, whereas unwillingness to receive psychological counselling was negatively associated with acceptance. By contrast, perceived risks, trust, and perceived anthropomorphism did not show significant direct effects. In moderation analyses, the influence of hedonic motivation on acceptance was stronger among male participants than among female participants. These findings provide implications for designing LLM-powered conversational agents for non-clinical mental health support by emphasising performance value and enjoyable user experiences.Practitioner Summary This study used survey data to investigate pre-adoption acceptance intentions towards LLM-powered agents for non-clinical mental health support. The results showed that perceived benefits were the primary driver of acceptance. Additionally, user characteristics such as gender and prior attitudes towards counselling significantly influenced initial acceptance intentions.