Science magazine announces that “all academic journals will use artificial intelligence to check fraudulent images” – GIGAZINE

Science, one of the most prestigious academic journals in the world, announced on January 4, 2024, that all of its academic journals are using “unauthorized images” as research findings, and we also announced that we will introduce artificial intelligence (AI) to automate the detection process.

Real photos in 2024 | Sciences
https://www.science.org/doi/10.1126/science.adn7530

All scientific journals will now perform AI-powered screening for image fraud Ars Technica
https://arstechnica.com/science/2024/01/all-science-journals-will-now-do-an-ai-powered-check-for-image-fraud/

Improvements in image processing technology and the shift to processes where everything from paper submission to publication is sometimes done digitally, have made it easier to commit “research fraud” by altering images submitted as evidence of research findings. This has been a problem for many years. Traditionally, after a research paper is submitted, it is reviewed by experts to examine the validity of the research content and results, but it is impossible to fully detect images that have been deliberately manipulated to mislead, and this can be difficult. If falsification of research results is indeed discovered, it will cause significant damage to the careers of experts who could not have discovered it through peer review.

To detect fraud in research papers, use the naked eye and memory to spot fraudulent papers that reuse and manipulate images.specializedIn some cases, this is helpful, but in many cases, Adobe Photoshop, which can enlarge, mirror, and overlay images, was used to scan the sheets. On the other hand, the development of AI tools to detect fraudulently manipulated images is progressing, and it was reported in October 2023 that tools that are much faster and more accurate than human labor have emerged.

Research results show that AI tools that detect “unauthorized images” in scientific papers are actually more accurate than humans – GIGAZINE

In response, Science is an image analysis tool that uses artificial intelligence.ProofIt announced that it would detect fake images in all six magazines published by Science.

Proofig analyzes images and generates reports that show “signs of processing” such as overlaps with historical data, rotations, scale distortions, and splices. Depending on the research, a “rotated image of something” may be published as a legitimate image, so the paper editor manually checks the report detected by the AI, and there is a problem with the report detected by the AI. Decide whether it's a thing or not. According to Science, after several months of experimenting with Proofig, they found clear evidence that it can detect problematic images that contain images that have been doctored before publication.

In response to the Science announcement, technology media company Ars Technica said: “Proofig can detect problems with a certain degree of accuracy, and it is desirable to detect problems before publishing a research paper. However, it is important to emphasize that not all AI-based detection systems Something that can be completed.” For example, Proofig detects data duplication by checking it against a database when data is plagiarized from previous research papers. However, if the data in a non-commercial paper is dealt with in a fairly small area, there is a high probability that the database will not cover it. In addition, if the forgery was made from unpublished research results rather than the images used in published papers, it would not be detectable through data duplication.

Flag announced at the end of 2023EditorialIn 2024, he says, “2024 is a year that will bring with it many challenges.” By applying artificial intelligence to detect images in peer-reviewed papers, and increasing monitoring and careful handling of research errors, “we hope to build stronger trust and integrity in science over the coming year.”