PREVIEW

Study Warns AI Chatbots Often Provide Inaccurate Medical Information. A recent study finds that a significant share of medical information generated by AI chatbots is either inaccurate or incomplete, raising concerns about their reliability for public health use. The research, published in the scientific journal BMJ Open, evaluated several widely used AI systems by testing how they responded to common health-related questions posed by non-experts. Results showed that around half of the responses were problematic, with some containing potentially harmful guidance if followed without professional supervision. Chatbot performance varied depending on the topic, with more accurate answers in areas like vaccines and cancer, and weaker results in subjects such as nutrition and sports. Beyond accuracy, the quality of sources cited by chatbots was often poor, and the language used tended to be overly complex, limiting accessibility for the general public. The authors stress that these systems do not truly reason or evaluate evidence but instead generate responses based on statistical patterns in their training data. Experts point to the need to reassess how AI chatbots are used in healthcare communication, warning that their current limitations could pose risks if relied upon without professional oversight.
Instructions: THIS VIDEO MUST NOT BE EDITED FOR LENGTH TO COMBINE WITH OTHER CONTENT