Johannes Larson was out on a deer hunt in the forest when his cell phone rang. It was an email from the journal Forest Ecology and Management asking if he could be a reviewer for an article they had received.
Such requests come often. Serious scientific journals allow other researchers to review all articles before they are published, to ensure that the quality of the text and research is high enough. The system is called peer review (see fact box).
- Most often you think: Do I know this subject well enough to be able to review it?, says Johannes Larson, who researches forestry and soil science at the department of forest ecology and management at the Swedish University of Agriculture SLU in Umeå.
That was not the case this time.
- When I saw the title, I thought: That sounds interesting, that's exactly what I'm working on.
But when Johannes Larson read the article's summary, he recognized most of it from a study he himself published in the journal Scientific Reports a few years ago.
- It wasn't exactly my text. But it was an equally large landscape they surveyed, and they had the same test points and the like. It was a Chat GPT version of my own article, he says.
Fake, mass-produced and AI-generated research articles are becoming an increasing challenge for the research world. Last year, a group of researchers presented the Stockholm Declaration, a list of actions to address the problem they call "the biggest scientific crisis ever".
In the article Johannes Larson had to read, all figures were taken directly from his study. But the text had been changed.
- The spirit of each paragraph was similar to my article. Had I given the text to a colleague, I'm not sure they would have discovered this. It might have reacted to the fact that the text was a bit strange, but not detected that it was plagiarism.
He contacted the editor of the magazine, and received a quick response.
- "I can't believe it's true, but I actually can," she wrote, adding that they are getting more and more similar articles submitted.
Johannes Larson thought it was a one-time event. But less than three weeks later, he received another request from another editor at Forest Ecology and Management. This time too, the article was a plagiarism of one of Johannes Larson's own published studies in Scientific Reports. Just like last time, the images were the same, with modified texts and headlines.
- Possibly it was a little worse made. I had made a tree inventory in a table, and that is called a forest survey. But survey also means survey.
Instead of "forest survey" there was now "forest questionnaire" next to the table.
- So it was as if they had gone around and asked the trees how big they were!
For researchers looking for new positions or funding for their research, it is crucial to be able to show that you have had studies published in scientific journals. Submitting fake articles can thus be a way of cheating yourself to a better track record. According to publisher Elsevier, which publishes Forest Ecology and Management and nearly 3,000 other journals, the number of articles from rogue actors, such as so-called article factories, has increased.
"Our investigations show that irresponsible use of generative AI often occurs together with other underlying problems concerning the integrity of research. Therefore, we focus our work on identifying these underlying problems, regardless of whether AI has been involved or not," writes a representative of Elsevier to DN.
The publisher has invested in technology and established a team of more than 100 specialists in research integrity and publication ethics to find clearly flawed articles as early as possible in the process. Publishers, editors, reviewers and researchers have a shared responsibility to maintain trust in the review process and the integrity of the research, Elsevier believes.
"The peer review system is still fit for purpose, but it must be developed as the publishing environment changes. We continue to work with the research community on this," writes the publisher's representative.
Both articles Johannes Larson was offered to review had a single author, and both stated that they were affiliated with Northeast Forestry University, one of China's oldest forestry research and forestry education universities.
- There are many questions that arise connected to the sender and purpose. Does anyone benefit from their publication, or are they looking to fill the journal with rubbish?
- It is also worrying that it comes through the first filter at the journal. Bad luck for the sender that I was chosen as the reviewer. I have a good grasp of my own articles, says Johannes Larson.
