A Danish biologist was shocked to find that his name had been mentioned several times in scientific papers about millipedes – which were simply referring to non-existent papers.
therefore Retraction clock According to reports, Danish myriapodologist Henrik Enghoff’s Natural History Museum suspected that authors of papers in China and Africa had used OpenAI’s ChatGPT to find academic references — and it turns out, they were right.
The offending paper was initially removed by Preprints.org, a preprint archive run by academic publisher MDPI, in June after Enghoff’s colleague, David Richard Nash of the University of Copenhagen, alerted the editors to the errors.
Now, the paper has resurfaced online, misleading references and all, on a separate preprint platform called Research Square..
This is an infuriating new reality that should have any academic concerned about how AI tech could undermine the validity of scientific research.
Scientists have already shown that services like ChatGPT have an alarming tendency to “false” scholarly citations.
And it’s not just academic. Earlier this year, journalists here parents It turns out that AI chatbots have created entire articles with a byline of journalists who never wrote these non-existent pieces. Then there was the lawyer who infamously used ChatGPT in court cases while doing research for his client’s case, a decision that backfired spectacularly.
It’s a cat-and-mouse game that can be a major headache for researchers.
“We will immediately retract it and add the authors of this preprint to our blacklist,” Preprints.org editor Lloyd Shue told Nash in an email back in June.
Enghoff also said Retraction clock He was informed that the same preprint had been submitted to MDPI’s journal insectBut it was rejected.
Kahse Tadese Mawcha of Aksum University in Ethiopia, who was originally listed as a corresponding author on the offending preprint, joined the Danish newspaper. Weekend Back in July he used ChatGPT, and he later learned that the tool was “not recommended” for this task.
A new version of the paper, posted above Research Square, lists Mawcha only as the author. Some references to illusory documents allegedly written by Enghoff have since been deleted, although there are many other references to non-existent documents, such as Retraction clock Report
“What I personally find most worrying is that many of these references suggest that the pest effect of millipedes on crops is greater than it actually is,” says millipede expert Leif Moritz, a researcher at the Leibniz Institute for the Analysis of Biodiversity. Change, Bonn, Germany, said the blog. This is something that can lead to “wrong decisions by policy makers”.
Then why did the paper burst again? Editors at Research Square said Retraction clock That the “retraction notification feature is currently under repair,” means they are “manually catching this.”
Whether by mistake or negligence, this is an alarming new reality for scholars around the world. Powerful but flawed AI tools like ChatGPT are a bull in the China Shop of almost every field of knowledge, academia included — and it will be interesting to watch as everyone tries to find a new sense of balance.
“It undermines confidence in the scientific literature,” Enghoff said Retraction clock.
More on ChatGPT: Oh my gosh… kids are going to start writing college essays using chatgpt