Use of artificial intelligence in the preparation of scientific articles

We currently face a complex ethical scenario in scientific knowledge production. At the same time that the advancement of Artificial Intelligence (AI) tools offers significant opportunities to optimize scientific and academic work, there is an increase in scientific misconduct with its unethical use by some individuals. In some cases, researchers commit plagiarism by incorporating excerpts generated by AI into their scientific articles without the due attribution. One example was the retraction of an article whose introduction began with a standard sentence generated by ChatGPT in response to a command: “Certainly, here is a possible introduction for your topic”.¹ In addition, it has become common for students to use AI to carry out work and assignments without considering issues of plagiarism and scientific integrity. Given this challenging scenario, would it be possible to use AI tools as allies in scientific nursing production in an ethical and responsible manner? This editorial seeks to answer this question by exploring the concept of AI, its growth in the scientific community and to propose solutions for the ethical use of these tools in nursing research.

Artificial intelligence in scientific production

Although AI has gained notoriety recently, the term and its interdisciplinary scientific field of study have been recognized since 1956, and previous productions indicate that its theoretical concepts had already been explored for many years. Artificial intelligence is currently a dynamic and multifaceted field focused on the development of systems and technologies capable of simulating human cognitive functions, such as learning, reasoning, communication, decision-making and problem-solving autonomously. The aim of AI is to perform tasks through data interpretation, continuous learning and adaptation, and its performance is adjusted based on user feedback.²

A large language model was released and made available for free by Open AI only in November 2022, marking a breakthrough in the accessibility of AI technology for the use by the general public through an intuitive and user-friendly interface called ChatGPT. Its demo and research version, based on GPT-3.5, was made publicly available for experimentation, allowing reinforcement learning and human feedback, which would be incorporated into GPT-4.0.³

ChatGPT represented a significant advance for both science and society. Between 30 November/2022 and 3 June/2023, the interest in this tool grew by 112,740%, reaching 100 million active users in just two months and becoming the fastest-growing consumer application in history.³

It was not different in the scientific field. When searching for the term “ChatGPT”, 2,090 studies were published in PubMed/MEDLINE and 4,754 studies in Scopus (search on 20 January/2025) in year 2023 alone. The tool gained even more visibility when four articles listed it as a co-author, generating intense debates in the scientific community. Several news articles criticized this practice, defending the regulation of the use of AI in research and questioning whether ChatGPT could assume authorial responsibility in scientific articles.⁴-⁶

The Committee on Publication Ethics (COPE), the World Association of Medical Editors (WAME), and the International Committee of Medical Journal Editors (ICMJE) have stated that AI tools do not meet the criteria for authorship, as they cannot assume responsibility for the accuracy, integrity, and originality of the work submitted.

In addition, scientific publishers have taken a stand and revised their editorial policies, incorporating guidelines on the use of AI in the production of scientific articles (Chart 1). As a result of this editorial and ethical alignment, it became prohibited to list ChatGPT as a co-author. All AI-generated content must be rigorously revised, and authors are responsible for ensuring its compliance with scientific and ethical standards.

Chart 1
Description of what is and is not allowed by scientific journals in the use of artificial intelligence in scientific articles

What is allowed by publishers
Language improvements: AI tools can be used to improve the fluency of language use, readability, grammar, punctuation, style and tone of texts. This is especially useful for authors who do not write in their native language (Elsevier, Taylor & Francis, Springer Nature, Sage Publications); Image adjustments: AI tools can be used for simple changes such as brightness, contrast and color balance, as long as they do not compromise the integrity of the image (Elsevier); Exclusive use in images based on verifiable scientific data: AI tools can be used to create images based on scientific data that can be attributed, verified and revised. Use must adhere to ethical, copyright and terms of use restrictions (Springer Nature); Support with scientific literature: AI tools can be used for interactive search for scientific articles and literature selection (Taylor & Francis); Coding assistance: AI tools can assist in coding as long as the author reviews the generated content (Taylor & Francis); Authors’ responsibility: If authors use any AI tool to generate texts, they must ensure that the content is accurate and valid, that there is no plagiarism, that all relevant sources have been cited and all statements in the article reporting hypotheses, interpretations, results, conclusions, limitations, and implications of the study represent the authors’ own ideas (Sage Publications, Plos One, Taylor & Francis); Declaration: The use of AI tools must be declared in the manuscript, specifying: name of the tool, version, and purpose (Elsevier, Taylor & Francis, The Wiley, Sage Publications, Springer Nature). The publisher Plos One specifies that in addition to these descriptions, authors must evaluate the validity of the AI outputs; Grammar checking tools (e.g. Grammarly) or reference managers (e.g. Mendeley, Zotero) are allowed without the need for a declaration (Elsevier).
What is not allowed by publishers
AI cannot replace authors in their responsibilities (generating texts, original results, codes or primary data without robust methodology) nor receive authorship credit (Elsevier, Taylor & Francis, The Wiley, Sage Publications, Springer Nature) AI cannot create or alter images, figures, tables or graphs. The exception is when the use of AI is part of the study methodology, which is then detailed in the materials and methods section (Elsevier, Taylor & Francis, Sage Publications, Springer Nature) Editors and reviewers should not place manuscripts, supplementary materials or any other unpublished information in AI tools in order to protect confidentiality and intellectual property rights (Taylor & Francis)

Authors who have used any AI tool must respect the principles of transparency and honesty in research, and declare which AI tool was used in the manuscript and for what purpose. Each journal determines how and where this statement should be included.

Relevance of artificial intelligence in scientific production

As elucidated so far, the evolvement of AI tools brings complex challenges that justify careful contextualization. This does not mean that their use should be avoided; on the contrary. It is essential to recognize the opportunities provided by these tools, whose growing visibility stems from their potential to improve work processes in any sector of society. In the educational sector, for example, AI can benefit teachers and also contribute to student learning. Artificial intelligence can facilitate personalized learning experiences, develop educational opportunities based on each student’s strengths, interests, and limitations, provide individualized feedback, and assist teachers in developing methodologies, lesson plans, and automating tasks, from the simplest to the most complex.⁷,⁸

Therefore, AI literacy is essential for developing specific skills in researchers, educators, and other responsible stakeholders for the promotion of human protagonism in scientific development. Brazilian universities, research institutions, and funding agencies need to formulate policies and guidelines for the use of AI.⁹

Furthermore, students need to understand the importance of academic integrity, good research practices, as well as the consequences of scientific misconduct, and learn how to use AI tools ethically and responsibly. This includes understanding AI limitations and developing skills to create effective commands.⁸

Artificial intelligence tools for writing scientific articles

Although new AI tools with different purposes are constantly being developed, most are not aimed at research hence, not designed to follow scientific principles and rigor. However, interesting tools have been designed specifically for researchers and can be incorporated into the scientific production process in nursing. Chart 2 presents some of the main options, their purposes and information on free use.

Chart 2
Artificial intelligence tools that can be used in the development of scientific articles

Tool	Access link	Scientific purpose
Scispace	https://scispace.com/	Search for scientific articles, reading and understanding of articles, extraction of information, generation of citations, among others
Consensus	https://consensus.app/search/	AI-based search engine to find answers in scientific research
ChatPDF	https://www.chatpdf.com/pt	Reading and understanding of texts, charts, tables and images of scientific articles
ExplainPaper	https://www.explainpaper.com/	Reading and explanation of complex scientific articles. Allows for generating explanations at different levels of understanding: five-year-old children, primary school, secondary school, undergraduate, graduate and expert
Scite	https://scite.ai	Analysis of citations to assess the credibility of scientific articles, search for citations and development of arguments based on scientific articles
ChatGPT Copilot Perplexity Deepseek	https://chat.openai.com https://copilot.microsoft.com/chats/ https://www.perplexity.ai/ https://www.deepseek.com/	Exploration of ideas, structuring of texts, offer of suggestions for improvement, grammar review, translation, among other commands that can be requested
ResearchRabbit	https://researchrabbitapp.com/	Facilitation of the discovery of relevant scientific literature, mapping of articles and connection of similar citations. Allows for tracking authors, topics and publications
Evidence Hunt	https://evidencehunt.com/	Identification of scientific evidence that answers specific questions and can support data-based health decisions. It is useful for quick reviews
Elicit	https://elicit.com/	Helps in identifying scientific literature and extracting information from articles
Napkin	https://www.napkin.ai/	Transformation of selected texts into images or other visual resources

Artificial intelligence tools can assist in several stages of research, including the search for scientific literature, extraction and organization of data, article reading, and writing of scientific papers.

SciSpace and Consensus are two useful tools for searching scientific literature. Both operate based on a question formulated by the user: first, a question is prepared, and the AI searches scientific articles for answers, highlighting the exact excerpt of the study that answers the question. These tools expand the possibilities in preparing a scientific study, as the argument or idea to be presented in the article can be transformed into a question, and from this point, the articles that best answer this question are found, providing a solid basis for constructing the argument in the article.

Scispace and Elicit are tools that assist in extracting data and preparing scientific study reports. In these tools, it is possible to load an article or select one of those retrieved by SciSpace and define what information the AI should extract, such as the study’s objective, methodology, and limitations.

Tools such as SciSpace, ChatPDF and ExplainPaper stand out in strategic reading of scientific articles. In SciSpace, it is possible to interact with a chatbot that answers questions about the study. ChatPDF has a similar functionality, in addition to allowing the selection of excerpts from the article for detailed explanation, always indicating the exact location of the information in the study. It is also possible to obtain explanations about images, graphs, charts and tables. ExplainPaper has an interesting function of explaining an excerpt from the article based on choices of different levels of understanding: five-year-old child, primary school, secondary school, undergraduate, postgraduate and expert.

Tools of AI also offer support in the field of translation and scientific writing. ChatGPT, for example, can help with the translation of texts, the writing of academic content in other languages and grammatical review, contributing to textual fluidity and cohesion. However, AI must not be used to generate texts in their entirety, but rather to improve content already developed by the researcher.

The possibilities for using chatbots, such as ChatGPT, Deepseek and Perplexity are broad and depend on the user’s creativity, because AI-based chatbots are capable of talking to the user through text, audio, images and even video in a dynamic and interactive way. Users can solve any task that requires automation. In scientific research, for example, they can be used to suggest subtopics for a theoretical framework, indicate keywords for searches in databases or even for brainstorming on possible research topics and problems.

In this way, the aforementioned tools, as well as those presented in chart 2 can contribute significantly to scientific production in nursing by increasing efficiency and optimization, and allowing researchers to focus their efforts on the analytical and interpretative stages of studies. However, the use of AI must be aligned with good scientific research practices, avoiding plagiarism and excessive dependence on the content it generates, since only us, humans, can create original knowledge.

Therefore, the academic and scientific community is invited to develop and promote strategies for the integration of AI tools in a responsible and efficient manner, strengthening their use without compromising the quality and integrity of nursing research.

The virtual deuteragonist

In ancient Greek theater, the protagonist was the main character in the plays, equivalent to the central role in a film or series today. In addition to this, there was the deuteragonist, the second most important character in the story. This character could act as an ally by helping the protagonist in the development of his journey and in reaching his maximum potential, or as a rival by creating conflicts that, nevertheless, drove the narrative of the play. Following this same logic, we can reflect on the role of AI in our lives. It is already present in our society, and it is up to us to decide on how to use it: as an ally, enhancing our human skills and abilities, or as an obstacle to our personal, professional and academic growth. As Professor Stephen Hawking said, “AI could be the best or the worst thing that ever happened to humanity”.¹⁰ That decision depends on each one of us.

References

• ¹ Zhang M, Wu L, Yang T, Zhu B, Liu Y. RETRACTED: The three-dimensional porous mesh structure of Cu-based metal-organic-framework - Aramid cellulose separator enhances the electrochemical performance of lithium metal anode batteries. Surf Interfaces. 2024;46:104081.
• ² Haenlein M, Kaplan A. A brief history of artificial intelligence: on the past, present, and future of artificial intelligence. Calif Manag Rev. 2019;61(4):5-14.
• ³ Dempere J, Modugu K, Hesham A, Ramasamy LK. The impact of ChatGPT on higher education. ront Educ (Lausanne). 2023;8:8.
• ⁴ Stokel-Walker C. ChatGPT listed as author on research papers: many scientists disapprove. Nature. 2023;613(7945):620-1.
• ⁵ Thorp HH. ChatGPT is fun, but not an author. Science. 2023 ;379(6630):313.
• ⁶ Goto A, Katanoda K. Should we acknowledge ChatGPT as an author? J Epidemiol. 2023;33(7):333-4.
• ⁷ Sabzalieva E, Valentini A. ChatGPT e inteligência artificial na educação superior: guia de início rápido. Organização das Nações Unidas para a Educação, a Ciência e a Cultura; Instituto Internacional da UNESCO para a Educação Superior na América Latina e no Caribe (IESALC); 2023. 14p. Available from: https://unesdoc.unesco.org/ark:/48223/pf0000385146_por.locale=en
  » https://unesdoc.unesco.org/ark:/48223/pf0000385146_por.locale=en
• ⁸ Alasadi EA, Baiz CR. Generative AI in education and research: opportunities, concerns, and solutions. J Chem Educ. 2023;100(8):2965-71.
• ⁹ Sampaio RC, Sabbatini M, Limongi R. Guidelines for the ethical and responsible use of Generative Artificial Intelligence: a practical guide for researchers. Sao Paulo: Intercom, ABCP, Anpocs; 2024.
• ¹⁰ Hern A. Stephen Hawking: AI will be 'either best or worst thing' for humanity. The Guardian. 2016 [cited 2025 Jan 30]. Available from: https://www.theguardian.com/science/2016/oct/19/stephen-hawking-ai-best-or-worst-thing-for-humanity-cambridge
  » https://www.theguardian.com/science/2016/oct/19/stephen-hawking-ai-best-or-worst-thing-for-humanity-cambridge

Publication Dates

• Publication in this collection
  13 June 2025
• Date of issue
  2025