Communications of the ACM
AI Beats Human Sleuth at Finding Problematic Images in Research Papers
Artificial intelligence tools often can spot more duplicated images in papers than a human specialist can.
Credit: Sezeryadigar/Getty
A study by independent U.K. biologist Sholto David found that artificial intelligence (AI) can be used to identify image manipulation in research papers faster and more accurately than humans.
David spent several months visually reviewing more than 700 papers with relevant images published in Toxicology Reports from 2014 to 2023 and identified 63 papers with duplicate images.
David then used the AI tool Imagetwin to analyze the same papers. It flagged the 63 papers that David had identified and another 41 that he had missed.
The study ultimately identified duplicate images in about 16% of the papers analyzed.
Imagetwin, which is used by around 200 universities, publishers, and scientific societies, makes a "fingerprint" for every image in a paper, then looks for duplicates in the paper and in its database of past papers, which contains more than 25 million images.
From Nature
View Full Article
Abstracts Copyright © 2023 SmithBucklin, Washington, D.C., USA
No entries found