""The results from this study are both disturbing and not very surprising to me. This raises concerns not only for research integrity, but also for clinical workflows, insurance claims and legal contexts where imaging evidence is used.""
""41% raised concerns that AI scans might have infiltrated the data set. There was no difference based on the experience of the radiologists, who had between zero and 40 years of professional experience.""
""AI models such as ChatGPT and Gemini were only 57-85% accurate when teasing apart real and synthetic images, indicating that even advanced AI struggles with this task.""
Research indicates that most radiologists find it challenging to distinguish between real and AI-generated X-ray scans, with fewer than half successfully identifying synthetic images. A study involving 17 radiologists revealed that while they could identify AI-generated scans 75% of the time after being informed, concerns about the integrity of medical imaging data persist. The study also highlights that large language models struggle similarly, raising alarms about the implications for research integrity and clinical practices.
Read at Nature
Unable to calculate read time
Collection
[
|
...
]