Limitations of Chatbots
The study published in the journal BMJ Quality and Safety has warned that information obtained from AI chatbots can be incorrect or harmful. The study found that many times chatbots provide complex answers that require high educational skills to understand. Researchers have particularly emphasized that AI chatbots are being used by patients to get answers to their drug-related questions, but their answers are not always accurate. This can potentially have harmful effects on health.
Chatbots’ are reliable?
During the study, researchers asked a chatbot (Bing Copilot) questions about the top 50 drugs prescribed in the US. They then analyzed the answers provided by the chatbot. The results were alarming – only half of the answers were completely correct, while 26% of the answers did not match the reference data and 3% were completely inconsistent.What Should Patients Do?
The study highlighted that 42% of the answers had a slight or moderate risk of harm, while 22% of the cases had a high risk of harm. Researchers said that AI chatbots are often unable to understand the underlying purpose behind a patient’s question. Based on this study, researchers advise patients to consult their doctors or health professionals to get information about their drugs or health. AI chatbots may be technologically advanced, but their information is not always accurate, which can harm health.
