9

These days many people use GPT, but to what extent should it be used, and how come journals and many authors do not check the papers they publish?

I am having a really hard time publishing my papers with many reviewers asking me to keep modifying and updating my paper.In contrast, I saw a video of Professor Dr. Eyad Qunaibi showing some mistakes published in a well-known credible journal.

Here are 2 examples.

https://www.sciencedirect.com/science/article/abs/pii/S2468023024002402

enter image description here

and

https://www.sciencedirect.com/science/article/pii/S1930043324001298

enter image description here

I wonder how such mistakes pass multiple reviewers and authors and whether there are consequences for the journal rating for such things

4
  • 4
    reddit.com/r/ChatGPT/s/EkcpvIiPSX
    – JRN
    Commented Apr 2 at 2:51
  • 15
    "I am having a really hard time publishing my papers with many reviewers asking me to keep modifying and updating my paper." That shows that you have chosen a reputable journal and that the process is working. This is good to hear. Keep aiming high, and take advantage of the valuable feedback you are getting. Commented Apr 2 at 11:11
  • 7
    I'm a little surprised that such obvious ChatGPT plagiarism got through the review process. Commented Apr 2 at 13:46
  • 2
    @EJoshuaS-StandwithUkraine Don't be surprised. There are many similar cases, all published in paid journals belonging to exploratory groups like MDPI or Elsevier. It just shows how blatantly irrelevant most papers are, and how ridiculous is the current status of academic publishing. Commented Apr 2 at 22:09

2 Answers 2

11

but to what extent it should be used

A pretty good chunk of reputable journals at this point have guidelines on how (if at all) generative AI may be used in paper submissions.

how come journals and many authors do not check the papers they publish

Various combinations of:

  • Authors are lazy/in a rush/under a lot of pressure and don't properly proofread
  • Peer reviewers are lazy/in a rush/under a lot of pressure and miss the obvious ChatGPT-like statements, particularly if they skip reading the introduction
  • The journal is not a reputable one, and isn't getting reputable peer reviewers
  • The journal is a reputable one, and didn't get good peer reviewers for that particular paper
  • Hundreds of thousands of papers are published every year, and sometimes things fall through the cracks

and if there are consequences on the journal rating for such things

Since journals are mostly just rated by impact factor, the impact on the journal will likely be negligible if it's just one or two articles like this. It would need to be a common enough occurrence, and a big enough issue that people would need to stop submitting to their journal, or stop citing the journal, in order for them to feel an impact.

7

Your question about whether there should be "a penalty when such mistakes are published" seems to me to be slightly naive. It seems to imply that there is a central authority responsible for ensuring that a particular quality of academic publishing be maintained and who can mete out penalties to journal publishers that don't make the grade Unfortunately, the vast proliferation of predatory garbage journals (and I am not suggesting for a moment that either of the journals you have mentioned fall in that category) is proof that no such control exists.

At best (or worst depending on your viewpoint), the "penalty" that a publisher suffers is to have authors desist from publishing with them anymore. Again, the proliferation of low-quality journals shows that no matter how bad the publication, you can always find someone who will send their material there.

On the other hand, you as an individual, can choose not to send your work to a publisher who persistently falls below a standard that you believe is reasonable ... and indeed, that is what most academics do choose.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .