Study Highlights Public Health Risks from Inaccurate Chatbot Health Advice
A study published in the BMJ Open journal has found that popular chatbots often provide problematic health advice. Researchers from the UK, US, and Canada tested five chatbots, including Google Gemini, DeepSeek, Meta AI, ChatGPT, and Grok, on their ability to answer health-related queries. The study revealed that half of the responses were problematic, with Grok performing the worst. The chatbots were tested on questions related to cancer, vaccines, stem cells, nutrition, and athletic performance. Despite the inaccuracies, the chatbots often conveyed their responses with confidence, lacking necessary caveats or disclaimers.