The use of AI tools like Imagetwin is becoming more common as academic publishers grapple with the issue of image manipulation in scientific papers. While not all image manipulation is done with malicious intent, journals are keen to identify images with alterations that cross the line. Other publishers are also exploring AI tools, with some developing their own software for image checking. The ultimate goal is to incorporate AI tools into the paper-review process, similar to how software is used to scan text for plagiarism.
Key takeaways:
- Artificial intelligence tools are being used to detect image manipulation in scientific papers, with one study showing that AI can often spot more duplicated images than a human specialist.
- Imagetwin, an AI tool used by around 200 universities, publishers and scientific societies, compares images in a paper with more than 25 million images from other publications, making it the largest such database in the image-integrity world.
- While AI tools like Imagetwin are effective, they are not infallible and can miss duplications in images with low contrast. They are best used as a supplement to human expertise.
- The ultimate goal is to incorporate AI tools into the paper-review process, similar to how software is used to scan text for plagiarism. However, human expertise is still necessary to interpret the results.