Adrian Johnson
TLDR:
- Software like Imagetwin and Proofig are helping publishers and universities detect copied scientific images.
- These tools automate the process of spotting deceptive images, which was previously done manually.
Software such as Imagetwin and Proofig are revolutionizing the detection of bad science by helping researchers and publishers identify copied or doctored images in scientific papers. These tools, like Imagetwin that compares photos against a database of 51 million images, can flag problematic images within minutes, a task that used to require manual scrutiny. Image expert Elisabeth Bik, who has spent over a decade scrutinizing scientific images, has found red flags in thousands of papers that were later retracted or corrected. While the technology is not perfect and may miss some issues, it has shown promising results in improving the quality of published science.
The tools, available for a fee, have been widely adopted by major universities and scientific publishers. For example, Proofig AI has tripled the rejection rate for papers at the Journal of Clinical Investigation by detecting disqualifying duplications or errors in images that were missed during peer review. These tools offer a chance to avoid lengthy postpublication postmortems and improve the reliability of scientific research. However, researchers are also wary of the emergence of sophisticated algorithms capable of fabricating fake experimental images, a new challenge that tool developers are working to address. Overall, technology is playing a crucial role in accelerating the hunt for bad science and enhancing the credibility of published research.
