A little over a week ago I posted extracts from an article about the burgeoning publication of bogus science online. This reply mentioned some of the individual "sleuths" attempting to track down the fakes. healthunlocked.com/cllsuppo...
Now, the science publisher Nature describes how journals are employing both sleuths and AI to identify problem images in scientific papers
nature.com/articles/d41586-...
Elisabeth Bik, a scientific image sleuth in San Francisco, and her colleagues examined images in more than 20,000 papers published between 1995 and 2014. They found nearly 4% contained problematic figures, with an increase around 2003, when digital editing of photographs became possible.
In an effort to reduce publication of mishandled images, some journals, including the Journal of Cell Science, PLOS Biology and PLOS ONE, either require or ask that authors submit raw images in addition to the cropped or processed images in their figures.
Many publishers are also incorporating AI-based tools including ImageTwin, ImaCheck and Proofig into consistent or spot pre-publication checks.
Users report that AI-based systems definitely make it faster and easier to spot some kinds of image problems. The Journal of Clinical Investigation trialled Proofig from 2021–2022 and found that it tripled the proportion of manuscripts with potentially problematic images, from 1% to 3%2.
But they are less adept at spotting more complex manipulations, says Bik, or AI-generated fakery. The tools are “useful to detect mistakes and low-level integrity breaches, but that is but one small aspect of the bigger issue,” agrees Bernd Pulverer, chief editor of EMBO Reports. “The existing tools are at best showing the tip of an iceberg that may grow dramatically, and current approaches will soon be largely obsolete".
As the above article makes clear, not all problematic images in scientific papers are deliberately altered. And manipulated images are just one kind of manipulated evidence, which in turn are only part of the wider issue of fake science. But, call me a doomsayer, what this article illustrates is that despite the best efforts of trained consultants (sleuths), AI-driven software, and pre-publication strictures introduced by the journals, they are going to have a hard time preventing fake journalism from permeating the internet - in all fields, including science and medicine. "A hard time" implies higher costs for publishers and paywalls for any content that has been through a recognised verification process. Of course, a paywall can also be spoofed...
