Safety · The Decoder ·

AI chatbots reading X-rays can be dangerously confident even when they're wrong

AI chatbots reading X-rays can be dangerously confident even when they're wrong

The RadLE 2.0 benchmark evaluated whether AI models used in radiology can recognize when they should defer to human experts. Many models made incorrect findings with high confidence, while radiologists performed better, highlighting a need for improved uncertainty handling before autonomous diagnosis.

Read the full story at The Decoder →