Are chatbots acceptable in mental health promotion and care? A cross-sectional study of mental healthcare professionals

Although chatbots are increasingly used in mental health, Finnish professionals’ perceptions remain unknown. This cross-sectional study (n=784) applied a modified UTAUT model to explore their perceptions of chatbots and their willingness to recommend them to clients or patients. Data were collected using non-probability sampling and analyzed with confirmatory factor analysis, multinomial logistic regression, and descriptive statistics. Within the next five years, 51% of respondents were somewhat or very unlikely to recommend chatbots, whereas 49% were somewhat or very likely to do so. Perceived potential of chatbots was strongly associated with the recommendation intention (p< 0.001). Higher facilitating conditions (users’ skills, device availability; p=0.003) and higher age increased recommendation willingness (B=0.051, p=0.004) with the age effect concentrating on ‘somewhat likely’ category. Perceived risks statistically significantly hindered willingness to recommend a chatbot across all outcome categories (p=0.010 to <0.001). Comprehensibility (72%) and professionalism (69%) were perceived as the most important chatbot features. Chatbots were seen as particularly suitable for assistive tasks such as sending reminders (77%) and instructing exercises (70%). Mental health professionals need training on chatbots' benefits, risks, and underlying technologies. In the sensitive mental health context, chatbots' roles should be clearly delimited, supporting their deployment in assistive, low-risk tasks.