First of all, I want to mention that this paper used ChatGPT (AI) for the process of finding information, getting some summaries and verifying references.
So, what is the difference between AI-generated vs human-created content? There are several points, but the most important one is to keep in mind that AI has not been created to substitute human beings but to aid them, to assist them in order to keep the time for performing some of the tasks to the minimum (Lakhani, 2023; Losey, 2025).
AI uses in scientific writing
ChatGPT can alleviate the work of the authors from the very beginning. What is the first step? Generation of ideas. Certainly, AI can help us brainstorm, identify gaps and even expand the context. But still it lacks creativity, and this should always be kept in mind. Before any research, the final formulation of the research question is essential, and this cannot be done without thorough literature review. AI can assist here as well. It can even do the literature review itself (Biswas, 2023). But the caution should be exercised, as it has been proven that AI can, so to say, “hallucinate” and generate fake or inaccurate references (Else, 2023). So it's better to sort the papers yourself and just increase the speed of summarizing by means of artificial intelligence.
Can we allow ChatGPT to actually write a text instead of us? There are ethical issues to be taken into account (Liebrenz et al., 2023). It is safer to stick to using it as an assistant but not as a writer (Kitamura, 2023; Thorp, 2023). But what can be done usefully without jeopardizing the grey areas of ethics? So instead of asking to write, we might ask to polish, improve grammar and correct mistakes (Fatima et al., 2024). We can also ask for formatting or verifying the references, but it is better to avoid asking it to generate references (see above) (Bhattacharyya et al., 2023).
AI uses in publishing
As from the point of view of the persons from the other bank of the river, in journal publishing we can use the new tool first of all against its unethical and wrongful use. It is possible to detect plagiarism, improve fluency and grammar and check the references.
As the first steps when getting the new papers, the editors can use AI in order to do so-called scope matching – check whether a manuscript fits the journal’s aims and scope. Next are the technical checks – like verifying whether author guidelines in regard to word count, formatting and structure have been complied with. The grammar issues and incoherence can be determined as well.
AI can help us not only to detect plagiarism but even paraphrasing and translated duplication and, last but not least, self-plagiarism (Gao et al., 2023; Quidwai et al., 2023; Wahle et al., 2021). It can identify statistical errors and inconsistencies, image manipulation and missing ethical statements.
AI can help match manuscripts to potential reviewers based on their expertise (keywords and past publications), past performance (speed and quality) and even flag conflicts of interest (Farber, 2024; Hosseini and Horbach, 2023; Pluto Labs, 2024).
Notwithstanding the fact that nowadays AI can predict the likelihood of acceptance of the manuscript and future citations, the final decision should remain human.
Conclusion
As we have mentioned in our editorial (Ibragimova and Phagava, 2024), it is absolutely necessary to disclose the use of AI, pointing out at what stage and for what tasks it has been used. More open discussion is necessary so as to not convert the use of AI into the magical power available just for the “chosen.” Sometimes authors think that nobody will find out and let the AI write instead of them. But, citing the title of the paper by Thorp et al., “ChatGPT is fun, but not an author” (Thorp, 2023), we absolutely need to keep this in mind.
