LLM for Philosophical Counseling - Study 1 & 2

This study compares the effectiveness of philosophical counseling delivered by human counselors versus by large language models (LLMs), with participants kept unaware of the counselor’s true identity (human or AI). It focuses on three primary outcomes: perceived professionalism, client satisfaction, and perceived authenticity of the counselor. Beyond this comparison, the study investigates three potential mediating mechanisms (perceived empathy, help-seeking shame, and self-disclosure) to understand how they function differently across human and AI counseling contexts. Additionally, it examines the moderating role of perceived social support, testing specifically whether it has an impact on outcomes for individuals when they interact with LLMs. We expect that AI counselors will perform no worse than human counselors on these outcomes. The influence of counselor type on these outcomes may be mediated by variables such as perceived empathy, help-seeking shame, and self-disclosure, and may be moderated by participants' levels of perceived social support.