Recently, we discovered a potential problem with the use of AI-generated articles. We would like your advice on formulating a process to screen for AI-generated submissions.
In the first review of an article, our EiC found that the author of the manuscript had no prior history of publication, and was connected with an AI website (found during an internet search). The article was not on the topic of AI. For an author who has never published an article and is one of the organizers of an AI website, writing a non-AI related article is obviously unreasonable. Based on these, the EiC thought the manuscript should be rejected.
Therefore, based on discussion in the editorial office, we propose paying special attention to the following points when initially checking manuscripts in order to exclude such AI-generated articles:
There are no published articles by the author on Google Scholar, PubMed, etc.
The author has no official institutional information.
The author's mailbox is not an institutional mailbox.
The article has only one author and no other co-authors.
The manuscript is submitted by the author spontaneously, with no prior contact or invitation.
We have also developed a list of features of AI-generated work for our editorial office, and a short guide for authors to use when using AI and AI-assisted technologies.
Questions for the Forum:
- Are these checks acceptable good practice?
- Are there other measures and methods to exclude such articles for publishers?