Artificial research

Lonni Besançon, an assistant professor in data visualization at Linköping University in Sweden, is also a detective of sorts. A self-­described “opportunistic” sleuth, he spends his free time hunting down fraudulent research papers.

According to Besançon, a quick Google search can easily expose fake academic authors and institutions. Less obvious fabrications require more digging, and some expertise: “That would be like, you know, looking at the plausibility of p-values.”

The professor is one of a host of scientific integrity sleuths sounding the alarm on counterfeit research. The number of research articles retracted in 2023 hit an all-time high, with over 10,000 papers pulled for fraudulent practices. The fraud ranges from images recycled from previous papers to entirely fabricated datasets. The sham papers aren’t only annoying, they’re harmful to ­scientific fields and in some cases even ­dangerous. Such academic cheating may be driven in large part by the strong career pressure researchers face to publish frequently.

The number of papers retracted increased more than five-fold between 2013 and 2023, according to a Nature analysis. A 2012 study in Proceedings of the National Academy of Sciences found that two-thirds of retracted biomedical and life sciences papers were withdrawn due to misconduct, including fraud, duplicate publication, and plagiarism.

Besançon argues the body of fraudulent research is actually far greater than what’s being retracted. “Ultimately we’re just catching the easy-to-catch ones,” he said.

The majority of articles retracted in 2023, over 8,000, were published by Wiley subsidiary Hindawi. Acquired by Wiley in 2021, Hindawi recently published a white paper blaming the uptick in pulled papers on “paper mills,” third-party individuals or groups that offer researchers authorship of sham papers for a fee.

David Bimler, a retired psychology researcher in New Zealand, sniffs out paper mills under the pseudonym Smut Clyde. He said that in Hindawi’s case, the paper mills exploited the publisher’s periodic special issues. Such issues focus on a particular topic, rely on a guest editor to recruit article authors, and require little publisher oversight, making them easy targets for scammers to slip in sham research.

Wiley has closed four especially problematic journals and in December announced plans to retire the Hindawi brand name. Matthew Kissner, Wiley’s interim chief executive officer, told Nature he anticipates the paper retraction scandal will cost his company $35 million to $40 million in revenue.

Integrity sleuths believe fraudsters are primarily responding to the ­pressure to publish. Researchers are assessed based on three metrics: ­number of publications, number of citations, and the h-index, which roughly quantifies a researcher’s productivity and impact. Besançon said those metrics can incentivize unscrupulous scientists to produce “bogus stuff. … There’s no value in doing one very good paper, but there’s value in doing six good enough papers.”

Hiring a paper mill makes it easier for a dishonest scientist to churn out a bogus paper—it’s like the high school jock who hires a math nerd to do his homework. “People clearly have strong career requirements to have a paper,” said Bimler, who believes some researchers view paper mill use as the price to get ahead.

While researchers exposed for fraud may face serious consequences, the fraudulent papers themselves carry serious implications for the scientific community. Some researchers may continue to rely on sham research unwittingly, even after an article has been retracted. If, for example, an investigator is running a human clinical trial relying on false claims that a certain medication has health benefits, the consequences could be catastrophic. “Imagine you then give that [medication] to people in another study somewhere else to just reproduce the results, and then you end up killing people,” Besançon said. “That could happen.”

Besançon thinks integrity sleuths are getting better at detecting paper mill–generated research, though. His colleague Guillaume Cabanac has developed a software tool called the Problematic Paper Screener. It identifies “tortured phrases,” strangely worded expressions resulting from automated attempts to hide plagiarism. But Besançon is less optimistic about sleuths’ ability to catch researchers who use more sophisticated fabrication methods.

Besançon said integrity sleuths aren’t paid for their work and instead often receive legal threats—or even death threats. If the research community offered protections and compensation, he suggested, the number of researchers engaged in sleuthing—as well as the amount of time spent sleuthing—might rise to meet the surge in counterfeiting.

“There’s more and more people interested in doing this, but there’s also more and more threats for doing it.”