Research say since the advent of AI, there has been a surge in papers that look scientific but don’t hold up to scrutiny. — Photo: Leonie Asendorpf/dpa
LONDON: Easy access to artificial intelligence (AI) has made medical and health research less scientifically rigorous and has facilitated a "flood" of shoddy journal papers full of superficial analyses based on "cherry-picked" data, a new study reports.
According to the University of Surrey and University of Aberystwyth, leaning on AI leads to the "production of large numbers of formulaic single-factor analyses" when a broader approach would likely better assess the range of possible causes of diseases.
Resorting to AI for a leg-up or head-start often ends up with researchers "relating single predictors to specific health conditions," the team said in a paper published by the science journal PLOS Biology.
"We’ve seen a surge in papers that look scientific but don’t hold up to scrutiny," said Matt Spick of the University of Surrey, who described such output as "science fiction."
The growing reliance on and hyping-up of AI is making so-called paper mills - where high volumes of quantity-over-quality medical or scientific journal papers get churned out - more proficient. Such would-be researchers can try to "exploit AI-ready datasets" to ensure "end-to-end generation of very large numbers of manuscripts."
According to the University of Surrey, some of the papers assessed were found to have featured "cherry-picked narrow data subsets without justification," practices that are "raising concerns about poor research practice, including data dredging or changing research questions after seeing the results."
Having thorough peer reviews and getting statisticians more involved with medical research that is based on large health datasets can help stem the tide, according to the team.
Researchers need to work harder too and "use the full datasets available to them unless there’s a clear and well-explained reason to do otherwise."
The growing use of AI means science publishing needs "better guardrails," according to the University of Surrey's Anietie Aliu. – dpa
