A new guide to designing, reporting and appraising studies of AI Chatbots

Generative AI chatbots (e.g. ChatGPT) are increasingly used in healthcare but many of the studies evaluating them are poorly done. Dr Anthony Sunjaya from the International Centre for Future Health Systems (ICFHS) and the School of Population Health at the University of New South Wales (UNSW) as part of the CHART Collaborative of global AI researchers has developed and published a new reporting tool for studies evaluating healthcare advice provided by chatbots (e.g. ChatGPT, Gemini, Copilot, Claude, DeepSeek). By applying CHART, researchers ensure their reports are clear, informative, and permit others to build on their work. CHART is technology-agnostic, ensuring it remains useful as technology progresses from large language models (LLMs) to vision-language models (VLMs), other foundation models, and beyond.