In recent years, the use of artificial intelligence (AI) has brought significant changes across various fields, including academia. Writing scientific journals, which was once considered a lengthy and complicated process, can now be simplified by utilizing AI tools specifically designed to support the needs of writers.
AI tools such as Grammarly, Quillbot, and Zotero have assisted many academics in crafting clear sentences, finding relevant references, and compiling bibliographies according to standard formats. Moreover, tools like Turnitin are now equipped with technology to detect AI-generated text, providing additional protection against plagiarism.
Furthermore, recent developments indicate that AI can also be used for data analysis and the creation of complex visualizations, making it easier for writers to effectively explain their findings. With this approach, AI becomes not just a supportive tool, but also a strategic partner in producing high-quality scientific work.
However, despite the many conveniences AI offers, academics are still expected to uphold academic integrity and ensure that technology is used ethically. AI is not a substitute for human creativity, but rather a tool to optimize existing potential.
The closing sentence above mentions academic integrity without citing any specific rules, even though the regulations have been written and are accessible to you. The Committee on Publication Ethics (COPE) states that AI tools should not be listed as authors. The reason is not a matter of preference, but rather a requirement of authorship: AI tools cannot bear responsibility for the submitted manuscript, and since they are not legal subjects, they cannot declare any conflicts of interest or manage copyright and licensing agreements. Authors who use AI to draft manuscripts, create images, or collect and analyze data must explain in the methods section which tools were used and how they were employed, while remaining fully responsible for all content.
The International Committee of Medical Journal Editors (ICMJE) regulates the technical aspects in more detail. Disclosure is made at the time of manuscript submission and is written in two places simultaneously: the cover letter and the manuscript itself. The placement follows the type of usage. Writing assistance is included in the acknowledgments section, while data collection, analysis, and image creation are included in the methods section. The last point directly touches on the example mentioned earlier in this article, namely using AI for data analysis and visualization. ICMJE also adds a warning that is easily overlooked: AI outputs may sound convincing but can be incorrect, incomplete, or biased, so authors are required to check and edit the results.
One thing in the second paragraph of this article also needs clarification. AI text detectors and plagiarism checkers are two different things, and the former does not produce verdicts. Turnitin itself published its figures on June 14, 2023, through their Chief Product Officer: the false positive rate at the document level is below 1 percent for documents containing 20 percent or more AI-generated text, while at the sentence level it is around 4 percent. In the same writing, educators are asked to treat flagged sentences as areas that need to be investigated and to use them to start a conversation, not to draw conclusions. Therefore, referring to it as an additional protection against plagiarism exaggerates two aspects at once: what is being measured and how certain the results are.
The practical consequences are light, as long as they are prepared from the start. The tools mentioned in this article remain useful; only the administrative notes change. Before submitting a manuscript, first check the AI policy of the journal you are targeting, as the details of the disclosure format may vary between publishers. Then write down three things: which tools were used, for which tasks, and in which section this is explained. The rest cannot be delegated. The responsibility for every sentence, including those whose first drafts were written by machines, remains with the name listed on the manuscript.
Sources
- COPE, reviewed February 13, 2023 – Authorship and AI tools, position statement on AI tools that do not meet authorship criteria and disclosure obligations in the methods section
- ICMJE Recommendations, section II.A.4 – Artificial Intelligence (AI)-Assisted Technology, where to disclose in the cover letter, acknowledgments, and methods
- Turnitin, June 14, 2023 – Understanding the false positive rate for sentences of our AI writing detection capability, false positive rates at the document and sentence levels