AI Chatbots Provide Inconsistent and Complex Medical Information on Neonatal Hyperbilirubinemia, Study Finds
A recent web-based cross-sectional comparative study evaluated the quality, transparency, and readability of information provided by four mainstream AI chatbots (ChatGPT 5.5, Gemini 3.5 Flash, Perplexity, and Copilot) regarding neonatal hyperbilirubinemia. The study found significant differences in the quality and transparency of information across the chatbots. While Copilot and Perplexity generally performed better in terms of information quality, none achieved an 'excellent' level. ChatGPT and Gemini exhibited particularly limited source transparency. A critical finding was that all chatbots generated texts with readability indices exceeding the recommended sixth-grade level for patient education, making the information too difficult for general patient comprehension. This suggests that while chatbots can offer structured supplementary medical information, they currently fall short in providing easily understandable and fully transparent content for public health education.
