Conversational agents (CAs), powered by artificial intelligence, are rapidly becoming integrated into mental health support as accessible, anonymous, and non-judgmental tools. While technical development has advanced considerably, far less attention has been paid to the psychological determinants that drive individuals' willingness to engage with CA-based interventions. This presentation reports findings from a cross-sectional web survey of 735 psychology students that examined the role of epistemic trust, attachment styles, personality traits, and fear of intimacy in shaping preferences for CA-mediated therapy.
Using ordinal logistic regression models, we found that higher epistemic trust strongly predicted willingness to use CAs for mental health support, while fear of sharing personal information (a facet of fear of intimacy) exerted a differentiated effect across levels of willingness. Personality traits also played a role: detachment negatively predicted CA preference, whereas psychoticism showed a positive association. Being single was a significant demographic predictor of greater willingness to rely on CAs.
These findings have direct implications for applied psychology. First, they highlight how psychological vulnerabilities—such as difficulties with intimacy or reliance on epistemic trust—shape engagement with novel therapeutic technologies. Second, they underscore the need for ethical and tailored integration of AI in clinical contexts, ensuring that CA-based interventions are not used as substitutes for human care where interpersonal connection is crucial. Finally, the results open new avenues for applied psychological research, inviting practitioners to consider how individual differences can inform the responsible design, implementation, and regulation of digital mental health tools.
By situating conversational agents within established psychological constructs such as attachment, trust, and personality, this work demonstrates how applied psychology can contribute to bridging technological innovation with human-centred care. The presentation will discuss practical applications, potential risks of overreliance on AI, and strategies for safeguarding therapeutic values in the digital era.