Large language model (LLM)-based chatbots such as ChatGPT, Gemini, Copilot, and Perplexity are transforming healthcare communication by delivering real-time conversational responses. While their use is expanding across clinical and nonclinical settings, concerns remain regarding their accuracy, safety, and potential to exacerbate existing health inequities. This commentary synthesizes emerging research to examine the applications, limitations, and risks associated with conversational AI in healthcare, focusing particularly on underserved populations with limited digital or health literacy. These tools are increasingly used in clinical settings for tasks like triage, surgical support, and diagnostic support, and in nonclinical roles such as health education and translation. However, chatbots often produce inaccurate, generalized, or biased responses due to their reliance on user-generated prompts and publicly available training data. These limitations disproportionately impact vulnerable populations, further deepening digital and health inequities. Addressing these issues requires inclusive chatbot design, content regulation, and public education on prompt literacy to ensure equitable healthcare communication.
Jamal Uddin (2025) studied this question.