Research repository ArXiv has banned authors for a year if they allow AI to do all the work

ArXivA widely used open source database for previously published research, it is doing more to prevent the indiscriminate use of large linguistic samples in scientific papers.

Although papers are submitted to the site before they are peer-reviewed, arXiv (pronounced “archive”) has become one of the main ways of conducting research in fields such as computer science and mathematics, and the site has become more accessible. source of data on trends in scientific research.

ArXiv has already started to take steps to combat the high volume of low-quality, AI-generated papers, for example in the search for first-time references. get approved by a regular writer. And after being with Cornell for more than 20 years, the institution is becoming a non-profit, which should allow it earn more money to solve problems like AI slop.

In his latest move, Thomas Dietterich – chair of arXiv’s computer science section – has been sent Thursday that “if the information presented contains irrefutable evidence that the authors did not look at the results of the generation of LLM, this means that we cannot trust everything that is in the paper.”

This irrefutable evidence can include things like “observable testimonials” and comments to or from the LLM, Dietterich said. If such evidence is found, the authors of the paper will face “a one-year ban from arXiv and be subject to the requirement that arXiv’s content must be approved by a recognized peer-reviewed site.”

Note that this does not prohibit the use of LLMs, but insists that, as Dietterich said, authors take “full responsibility” for the content, “regardless of how the content is created.” So if researchers copy-paste “inappropriate, false, biased, inaccurate, inaccurate, misleading, or misleading language” directly from the LLM, then they are responsible for it.

Dieterich he told 404 Media that this will be the law of “one sentence”, but the administration must announce the case and the regional seats must verify the evidence before imposing the sentence. Authors may also appeal the decision.

Recent research by colleagues has found this fiction is on the rise in medical research, perhaps because of LLMs – although it is true, it is not only scientists who are being caught. using words generated by AI.