Scientific journals are facing a flood of articles created using generative AI. While such works were previously easy to identify through illogical errors or fabricated citations, current tools can generate credible texts that are almost indistinguishable from real research. Editorial offices and reviewers are concerned that the peer review system is being overwhelmed. This is reported by Ixbt.com reports .
Peter Degen, a staff member at the University of Zurich, noticed that his old article suddenly had hundreds of new citations. It turned out that his work is being used en masse in fake studies automatically generated based on open databases like Global Burden of Disease. Such articles search for statistical correlations and present them as scientific innovations.
Scientific Reports editor Matt Spike notes that modern AI systems have become much more sophisticated. They independently analyze data, propose hypotheses, and write complete articles enriched with charts. Researchers at Carnegie Mellon University have found that such systems sometimes invent data or use questionable analysis methods, but the final result looks very professional.
Problems in the academic system are exacerbating the situation. Universities and grant foundations still evaluate scientists by the number of publications, while publishers profit from article processing charges. According to a study in the journal Nature, scientists who actively use AI publish three times more articles on average.
This process calls into question the scientific community's ability to distinguish real scientific achievements from statistical noise. The shortage of reviewers and the acceleration of publication flows are becoming one of the main factors leading the academic system into a serious crisis.
Read “Zamin” on Telegram!
