A new study compared four search engines and seven AI chatbots, including ChatGPT and GPT-4, to answer 150 health-related questions. Large language models outperformed search engines but showed sensitivity to prompts and risks of misinformation.