The Use of Large Language Model Chatbots for the Promotion of Physical Activity: A Scoping Review

Abstract

Physical inactivity is a critical challenge to global public health. Large language model (LLM) chatbots represent a paradigm shift in conversational agents, offering unprecedented capabilities for dynamic, personalized dialogue. The purpose of this scoping review is to map use of LLM chatbots for physical activity promotion, characterizing their application, their features, and functions, and their facilitation of behavior change techniques, to identify trends and gaps in the literature. Systematic searches of seven databases were performed for studies published between January 2017 and June 2025. Twelve studies satisfied this review’s inclusion criteria. LLM chatbots demonstrate significant potential for overcoming limitations of pre-LLM chatbots, enabling flexible physical activity interventions and facilitating complex behavior change techniques. However, the field remains nascent, evidenced by the prevalence of small-scale, pilot/feasibility studies. Future research should prioritize robust longer term effectiveness trials. Safeguards such as human–artificial intelligence oversight and retrieval-augmented generation should also be implemented in future work, particularly for vulnerable populations.