SCIENCE SLOP: Large language models and chatbots are growing in popularity for both recreational use and, in some cases, serious research projects. However, one of the internet's most popular research platforms is looking to curb misuse by users who rely on low-quality chatbot prompts without properly verifying the final output.
The arXiv (pronounced "archive") team recently announced a significant update to its official code of conduct. The popular open-access repository of research papers awaiting peer review will now seek to deter AI-generated "slop" by enforcing stricter accountability rules, including a one-year ban for violations. The team said that using LLMs and generative AI tools is generally acceptable, but only if the final output is carefully reviewed and corrected to remove hallucinated or inaccurate content.
Thomas G. Dietterich, a distinguished professor at Oregon State University and member of arXiv's moderation team, recently outlined the changes on social media. He explained that the updated code of conduct "states that by signing your name as an author of a paper, each author takes full responsibility for all its contents, irrespective of how the contents were generated."
