“The entire structure of science could collapse if this is left unaddressed,” says physicist Luís Amaral. He is referring to the fake research being published in scientific, medical, and academic journals that was written by “paper mills,” which are also bribing editors, journals, and peer reviewers. New investigations have uncovered the scope of this practice, which is staggering.
Some background: The major universities, public and private, are “research institutions.” Unlike smaller liberal arts colleges, they are not geared towards teaching but research. The idea is that university faculty members will be pursuing research in their fields, making the scientific discoveries and building up the knowledge base our society depends on. Students will benefit, so the theory goes, by helping the researchers in their work and by learning how to conduct research themselves. Most graduate degrees, whether master’s or doctoral, are research degrees, requiring original research projects written up as a master’s thesis or doctoral dissertation and qualifying the holders to work as researchers in research institutions.
Research is especially important in science. According to the scientific method and its professional protocols, experiments must be replicable–that is, repeated by other researchers who find the same results–and it must be shared with other scientists by publication. That way, science can build on the work of other scientists. Thus, thousands of researchers in many different fields–medicine, cell biology, geneticists, radiologists, pharmacologists, and more–are studying cancer. Their findings, however small they may appear, are stored and made accessible in scientific journals and they may someday come together into a cure for that disease.
Publish or perish: The scientific and academic profession thus require their members to do research and to publish the results. This has resulted in pressure to “publish or perish.” If you publish a lot, you are likely to get tenure in your university, with all the high salaries, job security, status, and other perks that go along with that. If you don’t publish–that is, if you either don’t do research or you can’t find a journal to publish it–you will lose your tenure track job.
Science has done amazing things, and yet scientists are human beings. They may not believe in the first three chapters of Genesis, but they are fallen creatures, susceptible to temptation and sin.
Paper Mills: We have blogged about the replication crisis, scholarly plagiarism, and falsified data. These are bad and damaging enough. Such violations of professional standards and ethics are usually committed by individuals acting on their own.
But, you know those term paper sites that operate openly and advertise online that students can use to buy research papers and other assignments to turn in as their own? Well, some of their professors are using that kind of service too.
“Paper mills” will generate research for you, making up data and writing it up for publication in a scientific journal. Not only that, some paper mills will create the scientific journal to publish your paper and that of their other customers. Some of them will provide the service of bribing editors and peer-reviewers of legitimate journals.
The existence of these paper mills has been known for a number of years, but we are now learning how much they are being used and how they are corrupting the scientific world.
The Wall Street Journal recently published a story by Nidhi Subbaraman entitled Scientific Journals Can’t Keep Up With Flood of Fake Papers. It reports on the findings of a large-scale study of published research conducted by the Proceedings of the National Academy of Science (PNAS), which identified work from paper mills by multiple tell-tale signs, such as the same photographs and illustrations appearing in many different articles and various language quirks.
The study found some 30,000 published articles in a wide array of legitimate journals from paper mills. It also found that certain editors and certain journals published more of them than others, suggesting collusion between the fraudsters and the publishers.
A separate investigation reported in Science found that paper mills were bribing editors, offering as much as $20,000 to publish the fake research. Another tactic is to get their paid agents appointed to journal publication boards.
The respected journal Nature has also been studying the issue of scientific fraud, which it describes as occurring on an “industrial” scale. In its 2021 article The Fight Against Fake-Paper Factories that Churn out Sham Science, it notes that the paper mills have been traced to Russia, Iran, and–especially–China. Many of the scientists who submit the phony research are also from China, where medical doctors also labor under the “publish of perish” system. The PNAS study cites a study that found that 46.7% of medical residents in southwest China admitted buying or selling papers.
This gets us to the fields where fake papers are most common. Not so much the social sciences (which have their own problems with problematic research), or the humanities (which would seem to be easier to fake). But the hard sciences. Specifically, medicine. Says another story on the topic, “Analysts’ data shows that fields related to cancer and medicine are particularly hard hit, while areas like philosophy and art are less affected.”
Cancer! PNAS data shows that genetics and cancer research are the fields most impacted by the paper mills. Which throws a big wrench into the hope that so much cancer research will come together into a cure for that disease!
The Impact on Science:
The PNAS report concludes, in part (my bolds & paragraphinc):
The trends we expose forecast serious risks ahead for the scientific enterprise. Large groups of editors and authors appear to have cooperated to facilitate publishing fraud (Fig. 1). Networks of linked fraudulent articles suggest industrial scale of production (Fig. 2). Organizations selling contract cheating services anticipate and counter deindexing and other interventions by literature aggregators (Fig. 3). The literature in some fields may have already been irreparably damaged by fraud (Fig. 4). Finally, the scale of activity in the enterprise of scientific fraud already exceeds the scope of current punitive measures designed to prevent fraud (Fig. 5).Currently implemented punitive measures are not addressing the tide of fraudulent science. First, papers published in deindexed journals remain a part of the record of the scientific literature in some literature aggregators (SI Appendix, Fig. S21). Second, retractions are still a relatively infrequent occurrence, far below what one would reasonably expect for clearly fraudulent papers (90). Only 8,589 of the 29,956 suspected paper mill products in our corpus that have a corresponding record in OpenAlex have been retracted (28.7%). Extrapolating from current trends, we estimate that only around 25% of suspected paper mill products will ever be retracted and that only around 10% of suspected paper mill products will ever reside in a deindexed journal (SI Appendix, Fig. S23).Collectively, these findings show that the integrity of the extant scientific record and of future science is being undermined through the shortcomings in the very systems through which scientists infer the trustworthiness of each other’s work.
The AI Effect: And, of course, the paper mills, though they got their start using corrupted human scientists to write the phony articles, are doubtlessly now using AI.
With this technology, scientists and academics who need to publish to avoid perishing can now dispense with the paper mills and just let an AI chatbot do their work for them. So we can expect the problem of fake science to accelerate.
But the effect on the sum total of human knowledge of industrial scale fakery plus artificial intelligence goes beyond the capacity of AI to generate more fakery.
The large language material that AI is trained on includes this fake science. Indeed, it includes the fake science generated by other AI programs. So as we depend more and more on AI, it will become more and more unreliable.
A few weeks ago we blogged about Ted Gioia’s contention that we are in the midst of a “knowledge system collapse.” Add to that what physicist Luís Amaral, one of the PNAS researchers, said, as we quoted before: “The entire structure of science could collapse if this is left unaddressed.”
Illustration: Cover of Science and Invention magazine (1927) via Picryl, Public Domain.
Ad-Free & Subscriber Log-In