Half of AI Health Advice Is Wrong—And Seems Just Right
60-second summary
Researchers in BMJ Open conduct a peer-reviewed audit, discovering nearly 50% of health responses from five major AI chatbots are problematic, featuring fabricated sources and confident delivery. This raises concerns about AI's reliability in healthcare. The industry implication is that AI health advice may be untrustworthy, necessitating stricter regulation and verification processes.
A peer-reviewed audit in BMJ Open found that nearly 50% of health responses from five major AI chatbots were problematic, with fabricated sources and confident delivery.