“I Don’t Understand It, but Okay”: An Empirical Study of Mental Health Practitioners’ Readiness to Use Large Language Models
Johanna Creswell Báez, Bryan G. Victor, Camille Dysart, Lauri Goldkind
Abstract
Large Language Models (LLMs) offer possibilities for changing the mental health services landscape. Given the fast moving nature of digital developments, there are a paucity of policy and practice recommendations for practicing professionals in the field. This qualitative study used demonstrations of the LLM ChatGPT and a case vignette of a patient describing preference for an LLM over a human therapist to elicit the perceptions and thoughts of therapists perspectives on LLMs and their possibilities for enhancing the work of mental health treatment. A total of 21 first interviews and 8 follow-up interviews with mental health professionals. The team used a narrative approach to analyze the interview data, identifying three broad themes: knowledge about LLMs, beliefs about LLMs and emotional responses to LLMs. These findings can inform user perceptions and practical applications for mental health practitioners and offer the field a reflection on how therapists are viewing these technologies.