Sign up to save tools and stay up to date with the latest in AI
bg
bg
1

AI-generated images threaten science — here’s how researchers hope to spot them

Nov 11, 2024 - nature.com
The article discusses the growing concern among scientists and researchers about the use of generative artificial intelligence (AI) in creating fraudulent scientific data and images. The ease with which these tools can generate text, images, and data has led to fears of an increasingly untrustworthy scientific literature filled with fake figures and conclusions. There is a growing arms race among integrity specialists, publishers, and technology companies to develop AI tools that can detect deceptive, AI-generated elements in papers.

The article also highlights that while AI-generated text is already permitted by many journals under certain circumstances, the use of such tools for creating images or data is less likely to be viewed as acceptable. There are suspicions that data and images fabricated using generative AI are already widespread in the literature. The challenge lies in detecting these AI-produced images as they are often almost impossible to distinguish from real ones. However, companies like Imagetwin and Proofig are developing AI tools to detect integrity issues in scientific figures and weed out images created by generative AI.

Key takeaways:

  • Generative artificial intelligence (AI) is becoming a powerful tool for fraudsters in the scientific community, raising concerns about the integrity of scientific literature.
  • AI tools can create text, images, and data that are difficult for humans to spot, leading to an arms race between integrity specialists, publishers, and technology companies to develop AI tools for detecting deceptive, AI-generated elements in papers.
  • There is a growing suspicion that data and images fabricated using generative AI are already widespread in scientific literature, with paper mills potentially using AI tools to mass-produce manuscripts.
  • Companies like Imagetwin and Proofig are developing AI tools to detect integrity issues in scientific figures and weed out images created by generative AI. However, the reliability of these tools is yet to be fully established.
View Full Article

Comments (0)

Be the first to comment!