A study found AI chatbots often struggle with clinical reasoning and may provide misleading medical advice. Researchers evaluated 21 AI models and found they failed to generate accurate differential diagnoses in more than 80 percent of cases. The study also reported that AI systems sometimes provided confident but incorrect responses and missed key symptoms. Accuracy improved only when complete patient data was available, which is uncommon in real-world scenarios. Experts said AI chatbots can assist with general information but should not replace doctors in complex medical decision-making.
A study found AI chatbots often struggle with clinical reasoning and may provide misleading medical advice. Researchers evaluated 21 AI models and found they failed to generate accurate differential diagnoses in more than 80 percent of cases. The study also reported that AI systems sometimes provided confident but incorrect responses and missed key symptoms. Accuracy improved only when complete patient data was available, which is uncommon in real-world scenarios. Experts said AI chatbots can assist with general information but should not replace doctors in complex medical decision-making.