To be Healed or Hacked? - User-Centered Ethical Design for Embodied AI in Mental Health Care

Area of Science:

• Mental Health
• Artificial Intelligence
• Virtual Reality

Background:

• Global mental health disorders present a significant treatment gap.
• Technology-mediated psychotherapy using AI and VR is being explored to increase access to care.
• Limited research exists on stakeholder concerns and readiness for AI in psychotherapy.

Purpose of the Study:

• To investigate patients' ethical concerns regarding AI in psychotherapy.
• To implement user-centered design measures to foster self-disclosure and trust in an embodied AI therapist within VR.
• To assess the impact of data autonomy and transparency on patient self-disclosure and trust.

Main Methods:

• An online study (N=152) identified patient concerns, prioritizing data autonomy and transparency.
• An in-person VR study (N=90) compared the effects of manipulated data autonomy and transparency on self-disclosure and trust.
• Statistical analysis was used to evaluate the impact of these variables.

Main Results:

• Increased data autonomy significantly enhanced self-disclosure in AI-driven VR therapy.
• Data transparency did not yield a significant effect on self-disclosure or trust.
• Neither data autonomy nor transparency significantly impacted overall perceived trust, but women reported higher trust.

Conclusions:

• Patient data autonomy is a key factor in fostering self-disclosure in AI-powered mental health care.
• Design considerations for AI therapists in VR should prioritize data control.
• Further research is needed to understand gender differences in trust towards AI therapists.