Withdrawn AI-written preprint on millipedes resurfaces, causing alarm

A preprint about millipedes that was written using OpenAI’s chatbot ChatGPT is back online after being withdrawn for including made-up references, Retraction Watch has learned. 

The paper, fake references and all, is also under review by a journal specializing in tropical insects.

“This undermines trust in the scientific literature,” said Henrik Enghoff, a millipede researcher at the Natural History Museum of Denmark, in Copenhagen, who first spotted problems with the preprint, as we reported last month. 

He and others said the case had caused a stir among researchers studying millipedes and centipedes, collectively known as myriapods. 

“The myriapodological community is alert and closely following this matter,” said Carlos Martínez, a centipede taxonomist and a research associate at Senckenberg Society for Nature Research, in Frankfurt, Germany. “I have also taken a personal interest in stopping this spurious manuscript from polluting the scientific record.”

The paper first appeared on Preprints.org in June. It came to Enghoff’s attention after he received an email from ResearchGate notifying him that his work had been cited. The citations were botched, he found, and some papers he was supposed to have written didn’t exist.

“This was so obvious that it had to be caught,” Enghoff told Retraction Watch. “But there could be subtler cases where something is published that is completely fictitious but won’t be detected.”

After learning of the fake references, staff at the preprint server, suspecting the involvement of a chatbot, pulled down the article and blacklisted its authors. 

The manuscript had also been submitted to MDPI’s journal Insects, but had been rejected, Enghoff said a colleague told him.

According to a Danish newspaper following up on Retraction Watch’s reporting, the preprint’s corresponding author, Kahsay Tadesse Mawcha of Aksum University in Ethiopia, admitted to using ChatGPT to produce the manuscript.  

Mawcha reportedly explained that he wasn’t “good at using AI” and had since realized that ChatGPT was “not recommended” for writing scientific articles.

Then on August 25, Mawcha posted an updated version of the paper to the preprint server Research Square. Two days later, an email appeared in Enghoff’s inbox alerting him that his work had been cited. It was déjà vu: Mawcha, who now appeared as the sole author of the manuscript, had deleted the made-up papers by Enghoff in the reference list, but the Danish researcher’s work again was being cited for something off-topic, Enghoff said.

What’s more, the revised preprint’s reference list still mostly consisted of fictitious papers, according to Leif Moritz, who has studied myriapods at the Leibniz Institute for the Analysis of Biodiversity Change, in Bonn, Germany.

“What I personally find most worrying is that many of these made-up references suggest that millipedes have a more negative effect as pests on crops than is actually true,” Moritz told Retraction Watch. “This in turn might not only lead to a negative perception of these animals in the public but also to misguided decisions by policy makers.”

Mawcha did not respond to a request for comments.

Asked if Research Square was aware of the manuscript’s troubled history, Mark Brewin, an editorial specialist at the company, told us: 

Our retraction alert feature is currently in for repairs, so meanwhile we’re endeavoring to catch these manually. Under the circumstances, our editorial team will look into this case and take appropriate action on the preprint.

On August 30,  the preprint website said Mawcha’s paper, titled “From Beneficial Arthropods to Soil-Dwelling Organisms: A Review on Millipedes in Africa,” was being reviewed by the International Journal of Tropical Insect Science. But the following day, after we contacted Research Square, all mention of the journal had disappeared from the website.

Ritter Atoundem Guimapi, an editor at the journal, told Retraction Watch he and his colleagues were aware of the issues related to the manuscript and were “handling the case accordingly.”

Meanwhile, Martínez, the centipede expert at Senckenberg Society for Nature Research, said he had emailed Mawcha and asked him to stop trying to publish “his AI-generated manuscript.” He also contacted the International Journal of Tropical Insect Science, he said, as well as the African Association of Insect Scientists and Ethiopia’s embassy in Germany. 

Note: The interview with Enghoff was conducted in Danish and translated by the writer, a native speaker.

Like Retraction Watch? You can make a tax-deductible contribution to support our work, follow us on Twitter, like us on Facebook, add us to your RSS reader, or subscribe to our daily digest. If you find a retraction that’s not in our database, you can let us know here. For comments or feedback, email us at [email protected].