Fake AI images will cause headaches for journals
Briefly

Research shows that AI can produce histological images so realistic that they often go undetected by viewers. A study involving 816 students revealed that many struggled to differentiate between real and fake images, which poses significant challenges for researchers, editors, and peer reviewers. To combat potential misinformation, the authors suggest that journals require publication of raw data to verify image authenticity. Despite existing tools for detecting image manipulation, they fall short against AI-generated images, raising concerns about the reliability of published research as technology continues to evolve.
A study presented to students revealed that many struggled to distinguish between real and AI-generated histological images, indicating a need for caution.
Participants unfamiliar with histological images were correct only 55% of the time, highlighting the challenge of identifying fake images.
Study co-author Ralf Mrowka emphasized the surprisingly low effort needed to create convincing histology images and the ongoing challenges of detection.
As AI image generation technology advances, the ability to reliably detect such fabrications poses significant challenges for researchers and publishers.
Read at Nature
[
|
]