AI chatbots give misleading medical advice 50% of the time, study finds
This image was generated by AI and may not depict real events.
A new study found that AI chatbots give misleading medical advice about half the time, highlighting health risks associated with the technology. Researchers evaluated five popular platforms, including ChatGPT, and found that about 50% of responses were deemed problematic, with almost 20% being highly problematic.
Researchers from the US, Canada, and the UK studied five AI chatbots. They asked each chatbot 10 questions across five health categories. About 50% of the responses were problematic. The chatbots performed better on closed-ended questions and worse on open-ended ones. They also struggled with topics like stem cells and nutrition. The study was published in the medical journal BMJ Open.
This content was automatically generated and/or translated by AI. It may contain inaccuracies. Please refer to the original sources for verification.