This issue can be viewed from both perspectives. On one hand, journals expect high-quality submissions; yet, when manuscripts are submitted organically, they are often returned with observations about quality or minor issues that could easily be resolved with the help of an agent or AI tools. As a result, many authors feel compelled to take that route.

On the other hand, journals themselves frequently take inordinately long periods to process manuscripts, often exceeding reasonable timelines. In some cases, conferences accept papers but then fail to publish them, which is even more frustrating for researchers.

Therefore, accountability is required on both sides. Authors are expected to act ethically by submitting their work to only one journal at a time. Equally, journals must uphold their responsibility by ensuring serious and timely scrutiny of submissions.

Finally, just as plagiarism is grounds for rejection, manuscripts with excessive AI-generated content may also be rejected. This balance ensures fairness and quality in scholarly publishing.

Mr. Kiran Kumar, you’ve raised important points. Like many of us, I’m on both sides of the process—as an author and as a reviewer. Recently I’ve noticed a concerning trend: a significant portion of submissions don’t even meet the basic scientific pillars that, as an author, I would never dare to submit. From my experience, around 3–4 out of every 10 articles I review fall into this category. This makes me wonder—what will happen if this proportion keeps increasing in the coming years?

Lee Wilson

Regarding your comment, I have also noticed that many authors are not putting the necessary care into aligning their submissions with the level of sophistication that a scientific article deserves. In my view, one of the main issues is not simply the style or structure, but whether the manuscript has a well-defined focal point that is consistently described in relation to a guiding research question and objective. When an article fails to connect its theoretical framework to its purpose, it often gives the impression of imprecision, as if the work is scattered rather than building toward a coherent contribution.

And to directly answer your question: the problems I’ve seen in recent submissions are mainly related to structure and to the scientific protocols that should guide a solid introduction, research method, development, and conclusion.

How to identify low-quality manuscripts and reject them at the point of submission, reducing the burden on the publishing process...

Meeting the challenges posed by mass-produced manuscripts and click-data science

"The combination of open-access datasets, machine learning workflows, increased computing capacity, and generative artificial intelligence has effectively removed many of the rate-limiting steps in manuscript production. This has created an industry of click-data science and a flood of low-quality manuscripts based on large health datasets such as the US National Health and Nutrition Examination Survey, the UK Biobank, and the US FDA Adverse Event Reporting System. These papers often employ statistically appropriate methods and real data, but introduce misleading results and false discoveries to the literature. Here, we offer suggestions for editors on how to identify such manuscripts and reject them at the point of submission, reducing the burden on the publishing process."

https://ese.arphahub.com/article/165043/

Unfortunatly, this is happening right now. Editorial personal should ve running an AI detecting test software to every submitted manuscript to detect it before considering for review

Ljubomir Jacić, thank you very much for sharing your insights. Here, we can see a legitimate and organized concern for outlining responses to this problem. Let's see if this impression we're having now diminishes in time. The material states that scientific production should focus more on the "why" than the "how," given the vast amount of secondary data we have available, which are used in a way that combines established methods.

Alejandro Rodriguez-Garcia, I truly agree that editors should conduct a preliminary analysis. I believe that Generative Artificial Intelligence resources can even be employed, as long as the information is coordinated through a structured and logical construction, resulting in high-level technical texts. What I've noticed is that people produce very low-quality texts, resulting in manuscripts incorporating all the jargon we're all too familiar with, which is produced by AIs.

I am a multiple Q1 and Q2 reviewer and writer for scientific health-related journals. IA is used to perform, for example, meta-analysis, which is the highest level of evidence analysis in medicine, and many authors use IA to describe, discuss, and even conclude on terms, concepts, and affirmations, among other crucial text content. This results in concepts, definitions, and conclusions that are irrelevant, repetitive, and wordy, lacking a strong, clear, and definitive message. This is more pronounced with narrative reviews on a concept, and the reading becomes tediously repetitive, boring, and useless.