Chatbots Produce Longer, More Emotional Health Responses

A 2026 JMIR study by Burns et al. compared emotional content, reading level, and disclaimer use for 100 physician-answered telehealth questions posed to OpenAI's ChatGPT and Google's Gemini on May 18–19, 2025. Researchers found chatbot replies were significantly longer, harder to read, and exhibited a wider range of emotions—Gemini showed higher fear and more disclaimers—compared with physician responses. These differences affect patient communication and chatbot deployment.