health
Don't Ask ChatGPT for Medical Advice: Half of Health Answers Are Wrong
Artificial intelligence provides incorrect or misleading medical advice in nearly half of cases. A major new study published in the esteemed medical journal BMJ Open has found that popular chatbots can be very dangerous to health. Researchers warn that people should not rely on these systems because the answers are often inaccurate or incomplete.
TL;DR
- Popular AI chatbots provide incorrect or misleading medical advice in nearly 50% of cases.
- A study in BMJ Open tested platforms like ChatGPT, Gemini, and Meta AI.
- Approximately 20% of the advice given by these bots was deemed highly problematic and potentially life-threatening.
- Chatbots were only somewhat reliable for simple questions about vaccines and cancer.
- Their knowledge failed on more complex topics like nutrition and stem cells, leading to numerous errors.
- None of the analyzed systems provided accurate references for their claims.
- AI chatbots often sound confident and knowledgeable, but their medical advice can be flawed.
- Consulting a human doctor remains the only safe way to get accurate diagnoses and treatment.