The Harvard School of Public Policy publishes a Disinformation Review of peer-reviewed scientific articles and promises "reliable, unbiased research on the prevalence, spread, and impact of misinformation worldwide."
This week he reported that “Academic journals, archives and repositories are seeing an increasing number of questionable research articles clearly produced using artificial intelligence".
They are often created with widely available AI applications, most likely ChatGPT, and mimic scientific writing. Google Scholar easily locates and lists these questionable papers. Analysis of questionable scientific papers produced with GPT and found on Google Scholar shows that many are often about controversial topics that are prone to misinformation: the environment, health, and IT.
The enhanced potential for malicious manipulation of society's evidence, particularly in politically divisive areas, is a growing concern... The abundance of fabricated "studies" seeping into all areas of the research infrastructure threatens to overwhelm the scientific communication system and endangers the integrity of the scientific record.
A second risk lies in the increased likelihood that persuasive scientific-looking content is actually created fraudulently with artificial intelligence tools and optimized to be retrieved by publicly available academic search engines, particularly Google Scholar. However small this possibility may be, awareness of it can undermine the basis of trust in scientific knowledge and poses serious social risks.
"Our analysis shows that questionable and potentially misleading papers produced by GPTs permeate the research infrastructure and are likely to become a widespread phenomenon..." the article notes.
“The centrality of Google Scholar to the publicly accessible scientific infrastructure, as well as the lack of standards, transparency and accountability for inclusion criteria, has potentially serious implications for public trust in science. This is likely to exacerbate the already known exploitability of Google Scholar for data breach…”
https://doi.org/10.7910/DVN/WUVD8X
