Usability and Acceptability of a GenAI-VR Onboarding Concept for Mental Health Support: An Interview Study

Abstract

Limited access to mental health care, driven by clinician shortages and systemic barriers, highlights the need for scalable and accessible support solutions. The integration of generative artificial intelligence (GenAI) and virtual reality (VR) offers a promising direction by enabling personalized, immersive, and continuously available mental health interventions. This study presents the design and formative evaluation of a low-fidelity GenAI-VR prototype aimed at supporting individuals with mild to moderate depressive symptoms. Rather than evaluating a full AI-VR therapy system, the prototype simulated only the onboarding and configuration components, specifically AI-guided avatar selection, environment choice, and the guided setup flow, using a paper-based interface. Ten participants who screened positive for at least mild depressive symptoms (PHQ-9 5) completed interaction tasks, followed by the Usefulness, Satisfaction, and Ease of Use (USE) questionnaire and semi-structured interviews. Results indicate that the configuration flow was perceived as easy to learn and use, with generally positive acceptance. Qualitative findings highlight both ongoing barriers in traditional care and user- identified design needs for AI- and VR-based systems, including personalization, anonymity, and culturally adaptable design, alongside concerns about privacy and the limits of AI emotional understanding. Overall, findings offer concrete design directions for researchers and developers building AI-assisted mental health tools, particularly for users blocked by cost, stigma, or geography. Future work should evaluate fully implemented systems to assess real-time interaction quality and sustained engagement.