Use of Large Language Models for Simulating Motivational Interviewing Counselling: A Scoping Review Protocol

Abstract Background Large Language Models (LLMs) such as GPT-4, PaLM, and LLaMA are increasingly applied in healthcare for tasks such as therapeutic dialogue and mental health support. Motivational interviewing (MI), a client-centred counselling method emphasizing empathy, partnership, evocation, and acceptance, has potential to be simulated by these models. However, questions remain about whether LLMs can replicate MI’s relational and emotional depth. This review aims to scope the evidence on the utilization and challenges of LLMs for delivering and simulating MI. Methods We will follow a five-step scoping review framework and report the review according to PRISMA-ScR guidelines. A comprehensive search of academic databases and grey literature will be conducted. We will include study that used LLMs for simulating MI among adults and all types of study designs. Discussion This will be the first scoping review to systematically explore AI-driven language models in relation to MI. The results will inform researchers, clinicians, and developers about the potential and limitations of LLMs in counselling, guiding ethical, safe, and evidence-based tool development. Trial registration: This protocol is registered on the Open Science Framework (OSF): https://doi.org/10.17605/OSF.IO/XEQYM