As a popular online research tool that gives users unlimited access to various interdisciplinary literature, Google Scholar is relied upon worldwide for conducting research and in-depth analysis. Sadly, a new study has revealed that Google Scholar has been exploited and used to legitimize over a hundred Junk, AI-generated scientific papers. This lapse could have serious implication on the public’s trust in science.
Researchers at the Swedish University of Agricultural Sciences recently discovered that AI-fabricated, essentially fake scientific papers are now on search engines, raising major concerns about scientific integrity. Björn Ekström, a Doctorate holder in Library and Information Science, explained that the availability of AI-generated studies increases the risk of “evidence hacking” which can make incorrect conclusions creep into society and influence critical decision-making processes.
Further investigation has also revealed that incorrect results obtained from this false information are now available across various research tools on the internet. As a result, millions of people are now exposed to misinformation on social media platforms and other outlets.
As AI-generated papers continue to proliferate, it may put additional pressure on the peer review process. Since generative AI can mass-produce fake scientific publications, reviewers will likely face a mounting number of submissions. This makes it more difficult to differentiate between AI-generated papers and genuine research, which may lead to inconsistent standards and reviewer burnout.
While explaining the implications of GPT fabricated, questionable studies, Professor of Library and Information Science Jutta Haider pointed out that there is a high probability of people drawing wrong conclusions from this incorrect information. The study highlighted that academic and scientific communication systems are under threat and made some recommendations to help identify potential AI-fabricated scientific papers. It also emphasized understanding how different approaches and “incentives relate to each other can help identify points and mechanisms of distributions.” Professor Jutta further highlighted the need for researchers to identify journals and publishers that present high-quality, peer-reviewed research in their academic databases.
The bottom line is, the mere fact that a research paper appears on Google Scholar does not guarantee its accuracy.