Scientific Publications Face Credibility Crisis

The quality and credibility of scientific publications have received increasing scrutiny. Findings from studies by Maria Ángeles Oviedo-García, PhD, from the Department of Business and Marketing at the University of Seville, Seville, Spain, highlight growing concerns about the integrity of published research. Insights from the journal Science and the US blog Retraction Watch revealed similar concerns regarding research integrity.

Artificial Intelligence (AI) Spurs Low-Quality Submissions

According to a report in Science, journals are inundated with low-quality contributions such as letters and comments generated by AI. Daniel Prevedello, MD, editor in chief of Neurosurgical Review, announced that the journal would temporarily stop accepting these submissions due to their poor quality.

Neurosurgical Review is not the only journal to experience low-quality submissions. In the journal Oral Oncology Reports (Elsevier), comments comprised 70% of the content, whereas in the International Journal of Surgery Open (Wolters Kluwer), they accounted for nearly half. In Neurosurgical Review, letters, comments, and editorials made up 58% of the total content from January to October 2024 compared with only 9% in the previous year.

This trend benefits authors by allowing them to inflate their publication lists with quickly produced contributions that bypass peer review. Publishers may also profit, as many charge fees to publish comments. Additionally, universities and research institutions find this type of content generation useful as more publications can enhance their reputation.

Concerns Over Peer Reviews

The troubling behavior described by Oviedo-García in the journal Scientometrics raises further doubts. An analysis of 263 peer reviews from 37 journals revealed that reviewers often used identical or very similar phrases in their evaluations, regardless of the content. In one case, the reviewer used the same wording in 52 reviews. This suggests that some reviewers read the studies that they are supposed to evaluate only superficially. Such practices can lead to valueless reviews and jeopardize the integrity of scientific literature. “Some other researchers will probably base their future research on these fake reports, which is frightening, especially when it comes to health and medicine,” Oviedo-García stated.

She suspects that the reviewers may have relied on templates to produce their reports quickly. This allowed them to list this work on their resumes for potential career advantages. Some reviewers have reportedly even “requested” the authors of the studies they reviewed to cite their own scientific work.

AI Complicates Peer Review

The process of research and publication has become increasingly challenging in recent years, and more standard and predatory journals allow anyone to publish their work for a fee. Roger W. Byard, MD, PhD, from The University of Adelaide, Adelaide, Australia, explained this trend in the journal Forensic Science, Medicine and Pathology. AI is increasingly being used to generate articles. At international conferences, experts have highlighted claims that AI can complete papers in just a few weeks and dissertations in less than a year. According to the authors of a letter in Critical Care, generative AI is infiltrating the peer review process.

Moreover, the peer review process can be bypassed by publishing research findings on online platforms (eg, preprint servers). Another issue is that some publications have hundreds of authors who can extend their publication list in this manner, even if their contribution to the publication is ambiguous or not substantial.

In a guest article for Laborjournal, Ulrich Dirnagl, MD, PhD, from the Charité — Universitätsmedizin Berlin, Berlin, Germany, emphasized that the scientific papers have become so complex that two or three experts often cannot thoroughly assess everything presented. The review process is time-consuming and can take several days for reviewers. Currently, very few people have time, especially because it is an unpaid and anonymous task. Dirnagl stated, “The self-correction of science no longer works as it claims.”

The old Russian saying, “Trust, but verify,” remains a timeless recommendation that is likely to stay relevant for years to come.

This story was translated from Univadis Germany using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication.