AI-generated disproportioned rat genitalia makes its way into peer-reviewed journal

The editors at the journal Frontiers in Cell and Developmental Biology have retracted a paper after it was pointed out to them by readers that supporting images had been generated improperly by an AI image generator. In their retraction, the editors report that the reason for the retraction was that "concerns were raised regarding the nature of its AI-generated figures."

In the article, which involved research surrounding stem cells in small mammals, the authors included images depicting rat anatomy that an AI system had clearly created. In one picture, a single rat appeared to have a penis and testicles that were larger than the rest of its body—not something that occurs in nature. Some of the accompanying text was also incomprehensible. Another image showed a rat cell that did not resemble the true structure of a rat cell.

The disproportioned images in the paper are likely to add to ongoing discussions in the science community surrounding the use of AI in generating text or imagery for use in technical papers—particularly those published in established journals.

In this case, it is not clear how such problematic images wound up in a peer-reviewed journal. The authors, a combined team from Hong Hui Hospital and Jiaotong University in China, did not try to hide the fact that they had used AI to create the images; they even credited Midjourney.

Some in the press have noted that Frontiers has a policy that allows for the use of AI-generated materials as long as their use is disclosed, which was the case in this instance. But the policy also notes that attempts must be made to fact-check anything produced by such systems, which clearly was not the case in this mix-up.

The editors at Frontiers initially posted a note on the paper claiming that the article had been corrected and that a new version would be published in short order. Not much later, the paper was retracted.

The mistakes made by the authors of the paper and the team at the journal that approved its publication are likely to be the first of many to come, though it is still not clear what changes will be required to prevent such mistakes from happening in the future.

© 2024 Science X Network