Navvya Jain

United We Care, India

Title: Blinded Assessment of User Preferences for Clinician-Patient Dialogue Styles: Comparing AI-Generated and Human Responses

---

Biography

Abstract

This study investigates user preferences for clinician-patient dialogue styles by conducting a blinded assessment comparing responses generated by AI-powered tools and human therapists. With the increasing adoption of Artificial Intelligence (AI) in mental health care, understanding user perceptions is critical to enhancing therapeutic approaches and integrating technology effectively. 500 global participants evaluated anonymized responses to common therapeutic questions, assessing them  across four key domains: comfort and empathy, perceived understanding, perceived helpfulness and overall preference without knowing the source.

The findings reveal nuanced preferences among participants, with AI-generated responses receiving significantly higher ratings across all evaluated domains. Interestingly, the qualitative feedback suggested that a portion of participants highlighted the potential of a hybrid model that combines AI and human therapists, leveraging the strengths of both. 

An important insight from the study was the influence of stigma associated with the term "AI." Many might express initial hesitation about engaging with AI-based responses, likely shaped by societal biases or misconceptions about AI's capabilities in therapeutic settings. However, when unaware of the response source, participants frequently rated AI-generated responses more favorably, indicating that such reservations may stem primarily from labeling “AI” rather than actual experience.

This research underscores the potential of AI as a complementary tool in mental health care, while emphasizing the irreplaceable role of human clinicians in providing nuanced, empathetic care. Future research should focus on refining AI systems to enhance their capacity for empathetic engagement and exploring hybrid models that integrate AI with human-led therapy to optimize patient outcomes.