The last few days have been quite agitated on the science side of Reddit and Twitter.
The culprit ? A peer-reviewed journal (meaning that an editorial board of subject experts review and evaluate submitted articles before accepting them for publication) decided that a paper containing blatant phoney AI-generated images could be published.
Frontiers in Cell and Developmental Biology is part of the Frontiers group, the 3rd most-cited and 6th largest research publisher and open science platform.
On 13th of February 2024, a paper titled “Cellular functions of spermatogonial stem cells in relation to JAK/STAT signaling pathway” (check this link for the PDF version) was published in the aforementioned journal.
Two days after, the first scientists were raising their concerns on Twitter and Reddit. The reason ? The article used absurd AI-generated images.
Take a look for yourself :
First of all, that’s a rat with a huge dong, yeah. And… 4 testicles ?
But it just get worse the longer you look. Other than the obvious absurdity at first glance, let’s check the labels shall we ?
“iollotte sserotgomar cell,” “testtomcels,” and “dck”.
Mmh. Okay. I mean, at least the rat is correctly labelled, right ?
But come on, maybe that’s just the first figure and they forgot to check it ? Let’s see the second one.
Ah yes, Jak -> Jak -> Jak -> Jak -> Tat signaling. Very well-characterized and well-known pathway.
But I must admit the captions are the star of the shows : “signal bıidimg the recetein“, “Sinkecler“, “dimimeriom eme“, “Tramioncatiion of 2xℇpens“, “ↄ“, and “proprounization“
Well, the last one can’t be worse, can it ?
It’s worse. I have no words. What the fuck is that ?
As appalling as this is, we can find an explanation after a quick research, as the images are credited : the authors used Midjourney, a popular generative AI tool.
If you’re wondering if it’s allowed, here is what Frontier’s policies for authors say : generative AI is allowed, but it must be disclosed, and the results must be checked for factual accuracy.
“Specifically, the author is responsible for checking the factual accuracy of any content created by the generative AI technology. This includes, but is not limited to, any quotes, citations or references. Figures produced by or edited using a generative AI technology must be checked to ensure they accurately reflect the data presented in the manuscript.”
Well, that did not go so well, did it ?
As the images caused an uproar on social media, some scientists contacted Frontiers in order to voice their (legitimate) concerns.
Consequently, the paper was retracted on the 16th of February, just 3 days after publication.
Frontiers issued a statement on their website :
Frontiers statement concerning the article “Cellular functions of spermatogonial stem cells in relation to JAK/STAT signaling pathway”, published on 13 February 2024
Thanks to the crowdsourcing dynamic of open science, we promptly acted upon the community feedback on the AI-generated figures in the article “Cellular functions of spermatogonial stem cells in relation to JAK/STAT signaling pathway”, published on 13 February 2024. Frontiers has now retracted and removed the article from the databases to protect the integrity of the scientific record.
Our investigation revealed that one of the reviewers raised valid concerns about the figures and requested author revisions. The authors failed to respond to these requests. We are investigating how our processes failed to act on the lack of author compliance with the reviewers’ requirements. We sincerely apologize to the scientific community for this mistake and thank our readers who quickly brought this to our attention.
As you can see, they were quick to defend the reviewers and accuse the authors.
But keep in mind that a paper like that has to pass editorial AND peer reviewing. How could that happen ? One user on Reddit had a theory :
As this add a little more context, it does not change the fact that reviewers did not do their job completely. They got at the end of the text and just.. stopped there ?
Frontiers’s statement says that “one of the reviewers raised valid concerns about the figures and requested author revisions“.
One ? What do you mean ONE ? Every single reviewer should have said something. Also, apparently reviewers comments mean nothing ? Why review papers at all in that case ?
It’s important to specify now that reviewers are volunteers. The responsability of this screw-up must be put on the journal and the editor. They’re employed exactly for this reason and the paper should have never passed this step.
To recap :
- Authors used AI and did not check their figures for accuracy
- Editor AND reviewers (minus one) did not check the figures of the paper
- Despite one reviewer’s comment, the paper got published without modification
Result : public outrage and the images going viral in the science community
Reaction from Frontiers : surprised pikachu face
I mean, what did you expect ?
Last thought
As funny as it is in this absurd situation, it’s also worrying about the future. If these pictures passed, how can we be sure that others did not ?
Elisabeth Bik, a renowned microbiologist and scientific integrity consultant, talked about it on her blog “Science Integrity Digest” :
But the paper is actually a sad example of how scientific journals, editors, and peer reviewers can be naive – or possibly even in the loop – in terms of accepting and publishing AI-generated crap. These figures are clearly not scientifically correct, but if such botched illustrations can pass peer review so easily, more realistic-looking AI-generated figures have likely already infiltrated the scientific literature. Generative AI will do serious harm to the quality, trustworthiness, and value of scientific papers.
Elisabeth knows what she’s talking about, she works on detecting photo manipulation in scientific publications and already identified over 4,000 potential cases of improper research conduct, including 400 research papers published by authors in China from a research paper mill company. Go check her blog !
I do agree with everything Elisabeth said.
It’s scary, and as scientists it’s our duty to keep a eye on published papers, and never forget science’s motto : “Question everything“.
Leave a Reply