Artificial intelligence in the preparation of scientific articles

S.M. Aldoshin,^a V.P. Ananikov,^b I.P. Beletskaya,^c A.A. Berlin,^d L.B. Boinovich,^m V.I. Bukhtiyarov,^f A.A. Voshkin,^e Yu.G. Gorbunova,^g,e K.V. Grigorovich,^h U.M. Dzemilev,^b M.P. Egorov,^b I.L. Eremenko,^e V.K. Ivanov,^e S.N. Kalmykov,^c V.P. Kolotov,ⁱ A.L. Maximov,^j A.M. Muzafarov,^k B.F. Myasoedov,ⁱ O.G. Sinyashin,^l K.A. Solntsev,^h A.R. Khokhlov,^c A.Yu. Tsivadze,^m A.A. Yaroslavov,^c A.B. Yaroslavtsev^e

^a Federal Research Center of Problems of Chemical Physics and Medicinal Chemistry of the Russian Academy of Sciences, Chernogolovka, Moscow Region 142432, Russian Federation

^b N.D. Zelinsky Institute of Organic Chemistry of the Russian Academy of Sciences, Moscow 119991, Russian Federation

^c M.V. Lomonosov Moscow State University, Moscow 119991, Russian Federation

^d N.N. Semenov Federal Research Center for Chemical Physics of the Russian Academy of Sciences, Moscow 119991, Russian Federation

^e N.S. Kurnakov Institute of General and Inorganic Chemistry of the Russian Academy of Sciences, Moscow 119991, Russian Federation

^f Federal Research Center “G.K. Boreskov Institute of Catalysis of the Siberian Branch of the Russian Academy of Sciences”, Novosibirsk 630090, Russian Federation

^g Skolkovo Institute of Science and Technology, Moscow 121205, Russian Federation

^h A.A. Baikov Institute of Metallurgy and Materials Science of the Russian Academy of Sciences, Moscow 119991, Russian Federation

ⁱ V.I. Vernadsky Institute of Geochemistry and Analytical Chemistry of the Russian Academy of Sciences, Moscow 119991, Russian Federation

^j A.V. Topchiev Institute of Petrochemical Synthesis of the Russian Academy of Sciences, Moscow 119991, Russian Federation

^k A.N. Nesmeyanov Institute of Organoelement Compounds of the Russian Academy of Sciences, Moscow 119991, Russian Federation

^l A.E. Arbuzov Institute of Organic and Physical Chemistry, Federal Research Center “Kazan Scientific Center of the Russian Academy of Sciences”, Kazan 420088, Russian Federation

^m A.N. Frumkin Institute of Physical Chemistry and Electrochemistry of the Russian Academy of Sciences, Moscow 119071, Russian Federation

Abstract

The use of artificial intelligence (AI) tools to prepare the text of scientific publications has already become an integral part of research practice. Large language models help translate and edit texts, structure articles, prepare abstracts, adapt manuscripts to journal requirements, accelerate the initial analysis of the literature and perform a range of other useful functions. However, the practical usefulness of a tool does not alter the fundamental principle: a scientific article remains the work of a researcher who is responsible for it. AI cannot be recognised as a co-author and cannot take responsibility for the authenticity of data, the correctness of chemical nomenclature, the logical presentation of the material, the completeness of citations or the validity of conclusions. This article presents the position of the editors-in-chief and editorial board members of several Russian chemistry journals on the use of AI in preparing scientific publications. A set of rules, VALENCE, is proposed as a practical framework for governing the use of AI in preparing scientific articles, consistent with international publishing practice. The bibliography includes 42 references.

Keywords: artificial intelligence, large language models, scientific publications, chemistry journal, authorship, editorial ethics, AI use disclosure, VALENCE, textual originality, reliability.

1. Can AI be used in scientific writing?

Yes, it can. Moreover, the scientific community can no longer ignore the rapid adoption of these tools in everyday practice. Researchers write articles in several languages, work with large bodies of literature, prepare responses to reviewers, ensure consistent terminology and perform numerous supporting tasks every day. Although these tasks are not the essence of scientific research, they require considerable time and attention. Large language models can help with this work. To date, no sound reasons have been put forward for prohibiting authors from using a tool that can process large bodies of literature and experimental data, improve the logical organisation of an article and the clarity of its text, reduce language barriers and help authors bring their writing into line with the standards of international scientific communication more quickly.

However, allowing AI use does not mean changing the fundamental nature of preparing scientific articles. A scientific article is not judged by whether the author typed every sentence without digital assistance. Its quality is assessed by the scientific merit of the results, their reliability and reproducibility, the accuracy and clarity of the presentation, the originality of the approach, the integrity of citation practices and, more broadly, responsibility for the reported results and conclusions. Using AI to help an author make the text more precise and understandable and improve the presentation of the material is consistent with scientific ethics. In contrast, using AI to create the appearance of new knowledge, falsify experimental data, generate irrelevant references, conceal weak reasoning or evade responsibility is not assistance to the author but a violation of scientific ethics.

Our editorial position is that we do not prohibit the responsible use of AI in preparing texts. We oppose irresponsibility. We do not propose rejecting a manuscript solely on the grounds that its author used a language model. However, journals should not consider articles containing unreliable results or statements, irrelevant references, plagiarism, unacknowledged borrowing, data falsification or incorrect chemical terminology, or articles prepared using AI tools without appropriate disclosure.

2. Scope of this article

This article deals exclusively with the use of AI in preparing the text of a scientific article: language editing, translation, stylistic revision, manuscript organisation, drafting abstracts and lists of keywords, suggesting alternative titles, preparing questions for checking the text, assisting with the organisation of a literature review, checking manuscript formatting against journal requirements and similar tasks. These are the technical aspects of manuscript writing and editing.

This article does not address the use of AI to create or edit figures, graphical abstracts, schemes, photographs, micrographs, spectra, images or visual representations of experimental data, nor its use in analysing raw data, planning experiments, generating models, automated synthesis or making scientific decisions. These areas have their own rules and risks and require separate editorial policies and consideration. This document concerns text only. The use of AI by editors and peer reviewers of scientific articles is also beyond the scope of this discussion.

3. Why researchers use AI in scientific writing

In a chemistry article, the text is not merely a decorative addition to the results. It is the means of describing an experiment and presenting its results logically and correctly, setting out the reasoning, the reaction mechanism, the limitations of the method and the conclusions. Poor writing can seriously undermine, or even misrepresent, the best results. In this respect, AI tools can be useful not because they write in place of the researcher, but because they can help the researcher organise the material, view their own text from a fresh perspective and improve it where necessary.

Large language models are particularly useful for tasks that involve working with language but do not require scientific decisions to be made on the author's behalf. These tools can generate several alternatives for clearer wording, shorten unnecessarily lengthy text, translate passages while preserving terminology, identify repetition, suggest a logical sequence of sections, draft an abstract within a specified word limit and compile a list of questions for checking the text. For authors who are not native speakers of English, AI can reduce barriers to international communication and allow them to focus on the substance of their research, considerably accelerating the preparation of high-quality texts in a foreign language.

In a recent classification of AI applications in chemistry, this mode of use corresponds to the “AI assistant” level: ready-to-use tools automate routine tasks, including translation, drafting articles and abstracts, initial literature analysis, keyword generation and the preparation of documentation.¹ These tools allow researchers to devote their attention to substantive work, but they also require new ethical and methodological standards, including disclosure of AI use and verification of the reliability of the output.¹

Scientific articles reporting research in chemistry and materials science are particularly sensitive to the quality of the text and the completeness with which the results are presented. An error in a substance name, an incorrect formula, an inaccurate or incomplete description of reaction conditions, incorrect reporting of the solvent, temperature, reaction time, reagent purification methods or spectral data processing, or an incorrect description of an analytical procedure can alter the meaning of the work. The use of AI in preparing publications is therefore acceptable only as assistance to the author and requires care and thorough verification of the text. The experimental section must be checked especially carefully, because even a slight misrepresentation of the conditions, procedure or processing of the raw data can reduce the reproducibility of the research.

4. Not humans versus AI, but the proper use of a tool

Discussions of AI too often present a false dichotomy: either a human author or a machine. Such an opposition is unproductive for science. Chemistry has already passed through comparable stages: computers, electronic databases, laboratory information systems, spectral processing software, automated systems for chromatographic and spectroscopic analysis, laboratory instrument control systems and many other innovations have not made the experimental chemist redundant. They have opened up new possibilities and changed the chemical expertise required to maintain an appropriate level of reliability in research results.

A key point today can be expressed as follows: “AI will not replace chemists, but chemists who rapidly learn to use its capabilities will inevitably move ahead of those who do not.”² This applies equally to the writing of scientific articles. Our task as researchers and authors of scientific publications is not to set scientists and artificial intelligence against each other, but to learn to use this new, modern tool in a way that preserves, rather than weakens, author responsibility.

Recent studies describe digital chemistry not as an abstract process of “digitalisation”, but as a transition from individual tools to an intelligent scientific infrastructure: AI is becoming part of data analysis, materials design, laboratory automation, natural language processing and the identification of hidden patterns.^3-5 Even in these more complex scenarios, however, the methodological requirement remains unchanged: data, code, models, conclusions and text must remain verifiable. In manuscript preparation, this means that authors may use AI as a tool but cannot delegate responsibility for the scientific content to it.

5. International publishing practice

International publishing practice has established several common principles in recent years. The first is that AI and large language models cannot be authors or co-authors of scientific articles, because authorship requires the ability to take responsibility for the content, sign publishing agreements and accept responsibility for the accuracy, originality and integrity of the work. The second is that AI use must be disclosed in the manuscript or at submission, in accordance with the journal's requirements, if it influenced the preparation of the text or content. The third is that human authors must check and edit the output, including facts, references and conclusions. These rules appear, in various formulations, in the policies of Elsevier,^6,7 Nature Portfolio and Springer Nature,^8,9,10 ACS Publications,¹¹ the Royal Society of Chemistry,¹² Wiley¹³ and Taylor & Francis,¹⁴ as well as in the ICMJE recommendations^15,16 and the COPE position statement.¹⁷

Importantly, these documents do not simply impose a ban. Elsevier explicitly recognises the potential of generative AI to improve authors' efficiency, but requires a disclosure identifying the tool, the purpose of its use and the extent of author oversight; in Elsevier's suggested wording, ultimate responsibility remains with the authors.⁷ ACS requires AI tool use to be disclosed in the manuscript and specifically states that AI cannot be an author and therefore cannot take responsibility for the published work.¹¹ Taylor & Francis permits the responsible use of generative AI for developing ideas, language improvement, translation, interactive searching, classification of the literature and assistance with writing and debugging code, while prohibiting uses that replace the core responsibilities of researchers and authors.¹⁴ ICMJE states the same principle: humans are responsible for any material prepared with the assistance of AI tools and must carefully check the output, which may be inaccurate, incomplete or biased.^15,16

The similarity of these publishers' positions is important for authors of scientific publications. It shows that the issue is already being addressed in terms of transparency, author responsibility and verifiability, rather than fear of, or uncertainty about, a new technology. Authors should therefore not conceal their use of AI, but should read the journal's rules in advance, use AI tools within the permitted limits and properly disclose their use in preparing the manuscript or accompanying documents.

Editorial policies have been discussed in Nature through calls for transparency and the exclusion of AI from authorship,⁸ in Science through the firm position “ChatGPT is fun, but not an author”,¹⁸ and in JAMA through a discussion of the implications of a “nonhuman author” participating in manuscript writing for the integrity of scientific knowledge.¹⁹ COPE and ICMJE have likewise affirmed human responsibility and the ineligibility of AI for authorship.^15,16,17 Publishers' positions were systematically examined in a review of instructions for authors, which identified a general trend towards prohibiting AI authorship alongside varying approaches to the disclosure of AI tool use.²⁰

Specific examples discussed in chemistry and related fields include accelerating progress in chemical science through AI,²¹ using ChatGPT for text analysis and predicting the synthesis of metal-organic frameworks (MOFs),²² and scenarios for transforming materials science and chemistry with large language models (LLMs).²³ The literature describes applications of LLMs and autonomous agents in chemistry,²⁴ autonomous chemical research with large language models,²⁵ the design of reaction arrays using ChatGPT²⁶ and assessments of the capabilities of LLMs for solving chemistry problems.²⁷

The 20 most influential AI technologies for chemistry and seven broad areas of digital transformation have been identified;²⁸ a three-level classification of AI use as “assistant–analyst–researcher” has been proposed;¹ a pragmatic approach to adopting AI as a tool, without expecting it to replace the chemist, has been formulated;² the problems of “dormant” and “lost” data have been examined;²⁹ and the “AI for” (AI4) era has been described as a shift from AI as an end in itself to AI as a tool for scientific and technological development.³⁰

A recent review places these studies in the context of digital chemistry as a distinct field.³ The literature also discusses the use of generative AI in chemical engineering, transparency, the ethics of professional use and applicability in the chemical industry.^31-36

6. The VALENCE rules for scientific writing

In our editorial practice, we propose following the VALENCE principles (Table 1), which define the relationship between the tool, the author and integrity in scientific writing. The VALENCE rules do not require authors to refrain from using AI. They require authors to retain responsibility for the publication. This is the fundamental distinction between the proper use of a tool and the ethically questionable replacement of the author.

Table 1. The VALENCE principles (Verify, Acknowledge, Limit, Edit, No AI authorship, Cite & protect, Ensure standards).

^a The rules are given in English to facilitate their integration into current international publishing practice.

7. How to disclose AI use properly

An AI use disclosure must be explicit, clear and comprehensive. It must identify the tool, its version or the date it was accessed, the purpose of its use and the fact that the authors checked the output. If the target journal provides its own template, authors must use it. If no template is provided, we recommend the following wording:

During the preparation of this manuscript, the authors followed the VALENCE recommendations [doi: 10.59761/rcr5247] and used {tool name, version/date} for {specific purpose/part of manuscript}. After using the AI tools, the authors reviewed, edited, verified and approved the entire text. The AI tools were not used as a source of scientific conclusions and are not an author or co-author. The authors take full responsibility for the scientific content of the publication.

With the text in {..} replaced by the corresponding information. When details are disclosed elsewhere, a shorter version may be also applicable :

During the preparation of this manuscript, the authors followed the VALENCE recommendations for using AI tools [doi: 10.59761/rcr5247]. The authors take full responsibility for the scientific content of the publication.

Where no AI tools were used, authors may state: “No AI tools were used in preparing the text.” Standard tools for checking spelling, punctuation and technical formatting generally do not require a separate disclosure unless the journal specifies otherwise. However, if an LLM-based tool did more than correct typographical errors and also suggested new wording, reorganised sections, produced drafts or analysed the literature, its use should be described.

Where necessary, a more detailed description of AI use in the work can be provided in the Methods section or in a separate section. One example is the disclosure in the article “Top 20 influential AI-based technologies in chemistry”, which reports the use of ChatGPT-4 for English language correction and literature analysis, including the classification, clustering and grouping of studies. It also explicitly states that the content was checked and edited and that the author takes full responsibility for the publication.²⁸ This approach—openly acknowledging the tool while affirming human responsibility—should become standard practice.

The possible consequences of an inaccurate disclosure should be determined by the rules of the particular journal and the nature of the established violation. Editorial decisions should be based on the violation of scientific or publication standards itself, rather than on the fact that AI was used.

8. What is acceptable and what is not

It is acceptable to use AI to improve the logical presentation of results, translate text into another language, make initial stylistic revisions, prepare alternative abstracts and keyword lists, draft a preliminary article outline, identify possible repetition, suggest improvements to the logical transitions between sections, prepare questions for the author, make an initial comparison of different journals' requirements and draft a response to a reviewer. A model may also be asked to explain which passages appear unclear, where the line of reasoning breaks down or which terms need to be used consistently. In all these cases, AI acts as an assistant, not as a source of scientific statements. Moreover, a model's response is probabilistic and should be treated as a suggestion for critical evaluation by the author, not as the result of verification. If a proposed change affects the interpretation of data or a scientific conclusion, the decision must remain with the author.

It is unacceptable to include AI-suggested references in an article without independently reading the relevant sources. It is unacceptable to ask AI to formulate results that are absent from the experimental data. It is unacceptable to use a model to generate a “convincing” discussion if the author cannot support every statement with a source, an experiment or a calculation. It is unacceptable to conceal substantial AI use where journal rules require disclosure. It is unacceptable to present generated text as scientific results that the author has not obtained, does not understand and cannot substantiate.

In chemical research, it is especially important not to treat AI as an unquestionable reference source. A model may confuse substances, name a reagent incorrectly, suggest a nonexistent reference, misinterpret a reaction mechanism, generalise a particular result into a universal rule or make an excessively broad generalisation, overlook specific features of an experiment, confuse reaction conditions, misstate units of measurement or confidently produce statements that sound plausible but are not supported by experiment, among other errors. All chemical details must therefore be checked by a human against primary sources.

9. Risks and ethical standards

The main risk of using AI to prepare scientific manuscripts is not machine assistance itself, but the production of text that sounds convincing yet lacks adequate scientific support. Large language models have no professional conscience, memory of laboratory work, experience of failed experiments or responsibility to the reader. They generate a probable continuation of a text, not necessarily the truth. Their output may therefore contain AI “hallucinations”, including fabricated facts or sources, incorrect identification of causes, sources or authorship, outdated information and polished statements with no methodological substance.

A second risk is plagiarism and copyright infringement. AI may reproduce passages closely resembling previously published text or introduce ideas without proper justification or supporting evidence. Responsibility for this lies with the author, not the tool. When using AI, authors are responsible for citing sources correctly, obtaining permission to publish material and ensuring that the work contains no plagiarism.

A third problem is the weakening of the author's judgement. If an author accepts a machine-generated draft without reading it critically, the article may lose precision, the author's own reasoning and its connection to the actual experiment. AI can make the style more consistent, but it cannot replace the author's scientific judgement and reasoned argument. This is particularly dangerous in chemistry: polished prose may conceal uncertainty about a reaction mechanism, an unsupported explanation of an experiment, an incomplete description of a method, insufficient reproducibility or misinterpreted data.

A fourth constraint is confidentiality. AI tools should be selected with due regard to security policies, licence terms and corporate and institutional requirements. Caution is needed when uploading an unpublished manuscript to external AI services, including services offering preliminary “expert” assessment: authors must check the terms governing confidentiality and the use of uploaded data beforehand.

Broader ethical and social issues also deserve separate consideration: the origin of training data and the rights of their authors, the resource requirements and environmental footprint of large models, and unequal access among researchers to the most powerful tools. These issues are important but lie beyond the scope of this article, which concerns the use of AI in preparing the text of a scientific publication.

The ethical boundary concerning authorship is currently clear: AI must not be listed as a co-author of a scientific article. This is not discrimination against a tool, but protection of the meaning of authorship. An author is someone who understands the work, agrees with the final version, can explain all the decisions made, can correct errors and can be accountable to the scientific community. AI is currently unable to do this and is unlikely to be able to do so in the near future.

10. Detection, text analysis and textual originality

Checking text for borrowed material and plagiarism is necessary. Textual originality remains an important criterion for scientific publication, and text similarity analysis helps identify improper borrowing, self-plagiarism, quotations that have not been properly identified and technical errors in references. However, assessing whether a text was “written with AI assistance” is a fundamentally different task. Such an assessment alone cannot establish whether an author has acted with integrity.

Our editorial position is that authors should not be required to provide an AI text detector report, and such a report should not, on its own, justify rejection of a manuscript. An instructive example is OpenAI, a leading global AI developer, which withdrew its AI text classifier because of its low accuracy.³⁷ Independent studies also confirm the limited reliability and vulnerability of detectors, showing that the tools tested were not sufficiently accurate.³⁸ Fundamental difficulties in reliably detecting AI-generated text have been examined both theoretically and empirically.³⁹ The risk of false positives and the need for professional judgement, rather than an automated conclusion that a violation has occurred, have also been emphasised.^40,41

What editorial teams need, therefore, is transparency and responsibility, not a detector of whether a text is “machine-written”.⁴² If authors honestly disclose AI use, check facts, verify references, preserve textual originality and take responsibility for the content, the use of AI itself is not a violation. Conversely, if a text contains plagiarism, false or unreliable statements, or undisclosed use of an AI tool where disclosure is mandatory, the problem is not that the text “looks AI-written”, but that scientific standards and scientific ethics have been violated. This does not mean that editorial teams should abandon verification: where there are specific inconsistencies in the data, references, originality or internal logic of a manuscript, editors are entitled to request explanations and conduct further checks. The assessment should concern the specific violation, not a probabilistic estimate of whether the text is “machine-written”.

11. The principal criteria for a scientific article

The editorial assessment of a scientific article must remain grounded in scientific criteria. Accuracy requires facts, numerical values, substance names and experimental conditions to be reported faithfully, and spectral and other data, formulae, units of measurement, analytical methods, statistics and conclusions to be correct. Clarity enables readers to understand what was done, why and exactly how it was done, what was obtained and the limits of the findings. Originality means not only the absence of textual plagiarism, but also an independent scientific contribution. Reliability means that statements are supported by measurement data (factual evidence), reliable sources and scientific reasoning. Reproducibility requires methods to be described in sufficient detail for independent verification. Transparency means that readers have access to information about sources, limitations, conflicts of interest and the tools used. Responsibility means that authors are prepared to substantiate every significant statement.

The proper use of AI is consistent with these criteria. Indeed, it can help authors express their ideas more clearly, identify incomplete descriptions, detect inconsistencies in the structure of the text, reduce language barriers and prepare drafts more quickly. However, AI does not relieve authors of their duty to verify the text. Scientific writing may be prepared with digital assistance, but scientific responsibility cannot be diluted or automated.

12. A practical workflow for authors

Before starting work, authors must check the target journal's rules: what the editors consider to be AI use, when disclosure is required, whether AI may be used for language editing and how substantial use of these tools should be reported in the Methods or other sections. Authors should then identify tasks for which AI use is permitted, such as translation, improving clarity, shortening the text or preparing questions for checking their own work. The more directly a task relates to a scientific conclusion, the more stringent the requirements for documentation and verification.

During the work, it is useful to retain key prompts and responses, particularly if AI helped structure a literature review or group sources. AI should not be used to generate a bibliography without subsequent manual verification of DOIs and bibliographic details. Once the work is complete, authors must read and check the text from a reviewer's perspective: where facts require a reference, where the wording is too general, where inaccuracies may arise, where fluent AI-generated wording conceals a lack of substance, among other issues. For papers with many co-authors, it is useful to agree in advance on a procedure for recording substantial AI use; technical tracking tools can assist but do not replace author responsibility.

At the final stage, authors must prepare an AI use disclosure. This disclosure is intended to help readers understand where AI tools were used and to identify text prepared with AI assistance. It does not diminish the value of the work, but can increase trust in it.

13. Conclusion

Science does not benefit from fear of AI tools. It benefits from accuracy, honesty, verification and clarity. AI has already become part of researchers' working practices. The issue is not to eliminate or restrict AI use, but to ensure that authors remain responsible for their work.

We accept the proper use of AI in preparing texts and welcome tools that help make an article clearer, more precise and more accessible to the reader. However, we require authors to check every reference, every number, every chemical statement and every conclusion. We insist that AI use be disclosed, that AI not replace the author's scientific judgement and that it not become a co-author of the publication to whom responsibility can be shifted.

The quality of an article in a chemistry journal is defined not by the avoidance of AI tools, but by the absence of unreliable information, carelessness, plagiarism, unacknowledged borrowing and irresponsibility. If AI helps authors achieve greater accuracy, clarity, originality and reliability, it can be a useful tool. If its use moves authors away from these criteria, the researcher must stop using it. The final decision always rests with the researcher.

14. Disclosure of AI use

During the preparation of this manuscript, the authors followed the VALENCE recommendations and used AI systems for language and structural editing, drafting alternative wording for individual passages, improving the logical presentation, organising information and formatting the manuscript. After using the tool, the authors reviewed and edited the entire text, agreed on its content and approved it. The AI tool was not used as a source of scientific conclusions and is not an author or co-author of the work. The authors take full responsibility for the scientific content of the publication.

15. Author information

S.M. Aldoshin. Russian Chemical Bulletin.

V.P. Ananikov. Doklady Rossiiskoi Akademii Nauk. Khimiya, Nauki o Materialakh (Reports of the Russian Academy of Sciences. Chemistry and Materials Science).

I.P. Beletskaya. Russian Journal of Organic Chemistry.

A.A. Berlin. Khimicheskaya Fizika (Chemical Physics).

L.B. Boinovich. Colloid Journal.

V.I. Bukhtiyarov. Kinetics and Catalysis.

A.A. Voshkin. Theoretical Foundations of Chemical Engineering.

Yu.G. Gorbunova. Russian Journal of Physical Chemistry A.

K.V. Grigorovich. Russian Metallurgy (Metally).

U.M. Dzemilev. Solid Fuel Chemistry.

M.P. Egorov. Russian Chemical Reviews, Russian Chemical Bulletin, Mendeleev Communications.

I.L. Eremenko. Russian Journal of Coordination Chemistry.

V.K. Ivanov. Russian Journal of Inorganic Chemistry.

S.N. Kalmykov. Radiochemistry.

V.P. Kolotov. Journal of Analytical Chemistry.

A.L. Maximov. Russian Journal of Applied Chemistry, Petroleum Chemistry.

A.M. Muzafarov. INEOS OPEN.

B.F. Myasoedov. Radiochemistry.

O.G. Sinyashin. Russian Journal of General Chemistry.

K.A. Solntsev. Inorganic Materials, Physics and Chemistry of Materials Treatment.

A.R. Khokhlov. Polymer Science.

A.Yu. Tsivadze. Russian Journal of Physical Chemistry A, Russian Journal of Electrochemistry.

A.A. Yaroslavov. Polymer Science, Series A, B and C.

A.B. Yaroslavtsev. Membranes and Membrane Technologies.

16. List of abbreviations

17. References

1. V.P.Ananikov. Analitika, 15 (4), 246 (2025); https://doi.org/10.22184/2227-572X.2025.15.4.246.260

2. V.P.Ananikov. Chim. Oggi – Chem. Today, 42 (4), 14 (2024)

3. A.O.Terentyev. Bashk. Khim. Zh., 33 (1), 7 (2026); https://bcj.rusoil.net/files/slider/BCJ_1_2026/BCJ-1-2026.pdf

4. S.Bräse. Digit. Discov., 3, 1923 (2024); https://doi.org/10.1039/D4DD00130C

5. I.V.Ananyev. Analitika, 16 (1), 34 (2026); https://doi.org/10.22184/2227-572X.2026.16.1.34.54

6. Elsevier. Generative AI policies for journals; https://www.elsevier.com/about/policies-and-standards/generative-ai-policies-for-journals (accessed 21 August 2026)

7. Elsevier. The use of generative AI and AI-assisted technologies in writing for Elsevier; https://www.elsevier.com/about/policies-and-standards/the-use-of-generative-ai-and-ai-assisted-technologies-in-writing-for-elsevier (accessed 21 August 2026)

8. Nature, 613, 612 (2023); https://doi.org/10.1038/d41586-023-00191-1

9. Nature Portfolio. Artificial Intelligence (AI): editorial policies; https://www.nature.com/nature-portfolio/editorial-policies/ai (accessed 21 August 2026)

10. Springer Nature. Editorial policies: Artificial Intelligence; https://www.springernature.com/gp/policies/editorial-policies (accessed 21 August 2026)

11. ACS Publications. Artificial Intelligence (AI) Best Practices and Policies at ACS Publications; https://researcher-resources.acs.org/publish/aipolicy (accessed 21 August 2026)

12. Royal Society of Chemistry. Author responsibilities; https://www.rsc.org/publishing/journals/processes-and-policies/author-responsibilities (accessed 21 August 2026)

13. Wiley. Best Practice Guidelines on Publishing Ethics: AI Technology; https://authors.wiley.com/ethics-guidelines/index.php (accessed 21 August 2026)

14. Taylor & Francis. AI Policy; https://taylorandfrancis.com/our-policies/ai-policy/ (accessed 21 August 2026)

15. International Committee of Medical Journal Editors. Use of AI by Authors; https://www.icmje.org/recommendations/browse/artificial-intelligence/ai-use-by-authors.html (accessed 21 August 2026)

16. International Committee of Medical Journal Editors. Use of Artificial Intelligence in Publishing; https://www.icmje.org/recommendations/browse/artificial-intelligence/ (accessed 21 August 2026)

17. Committee on Publication Ethics (COPE). Authorship and AI tools. 2023; https://publicationethics.org/guidance/cope-position/authorship-and-ai-tools (accessed 21 August 2026)

18. H.H.Thorp. Science, 379, 313 (2023); https://doi.org/10.1126/science.adg7879

19. A.Flanagin, K.Bibbins-Domingo, M.Berkwits, S.L.Christiansen. JAMA, 329 (8), 637 (2023); https://doi.org/10.1001/jama.2023.1344

20. C.Ganjavi, M.B.Eppler, A.Pekcan, B.Biedermann, A.Abreu, G.S.Collins, I.S.Gill, G.E.Cacciamani. BMJ, 384, e077192 (2024); https://doi.org/10.1136/bmj-2023-077192

21. S.Back, A.Aspuru-Guzik, M.Ceriotti, G.Gryn’ova, B.Grzybowski, G.H.Gu, J.Hein, K.Hippalgaonkar, R.Hormázabal, Y.Jung, S.Kim, W.Y.Kim, S.M.Moosavi, J.Noh, C.Park, J.Schrier, P.Schwaller, K.Tsuda, T.Vegge, O.A.von Lilienfeld, A.Walsh. Digit. Discov., 3 (1), 23 (2024); https://doi.org/10.1039/D3DD00213F

22. Z.Zheng, O.Zhang, C.Borgs, J.T.Chayes, O.M.Yaghi. J. Am. Chem. Soc., 145, 18048 (2023); https://doi.org/10.1021/jacs.3c05819

23. K.M.Jablonka, Q.Ai, A.Al-Feghali, S.Badhwar, J.D.Bocarsly, A.M.Bran, S.Bringuier, L.C.Brinson, K.Choudhary, D.Circi, S.Cox, W.A.de Jong, M.L.Evans, N.Gastellu, J.Genzling, M.V.Gil, A.K.Gupta, Z.Hong, A.Imran, S.Kruschwitz, A.Labarre, J.Lála, T.Liu, S.Ma, S.Majumdar, G.W.Merz, N.Moitessier, E.Moubarak, B.Mouriño, B.Pelkie, M.Pieler, M.C.Ramos, B.Ranković, S.G.Rodriques, J.N.Sanders, P.Schwaller, M.Schwarting, J.Shi, B.Smit, B.E.Smith, J.Van Herck, C.Völker, L.Ward, S.Warren, B.Weiser, S.Zhang, X.Zhang, G.A.Zia, A.Scourtas, K.J.Schmidt, I.Foster, A.D.White, B.Blaiszik. Digit. Discov., 2 (5), 1233 (2023); https://doi.org/10.1039/D3DD00113J

24. M.C.Ramos, C.J.Collison, A.D.White. Chem. Sci., 16, 2514 (2025); https://doi.org/10.1039/D4SC03921A

25. D.A.Boiko, R.MacKnight, B.Kline, G.Gomes. Nature, 624, 570 (2023); https://doi.org/10.1038/s41586-023-06792-0

26. B.Mahjour, J.Hoffstadt, T.Cernak. Org. Process Res. Dev., 27, 1510 (2023); https://doi.org/10.1021/acs.oprd.3c00186

27. T.Guo, K.Guo, B.Nan, Z.Liang, Z.Guo, N.V.Chawla, O.Wiest, X.Zhang. What Can Large Language Models Do in Chemistry? A Comprehensive Benchmark on Eight Tasks. arXiv, 2305.18365 (2023); https://doi.org/10.48550/arXiv.2305.18365

28. V.P.Ananikov. Artif. Intell. Chem., 2 (2), 100075 (2024); https://doi.org/10.1016/j.aichem.2024.100075

29. V.P.Ananikov. Vestn. RAN, 95 (4), 35 (2025); https://www.elibrary.ru/item.asp?id=82376411

30. V.P.Ananikov. Additiv. Tekhnologii, 3, 10 (2025)

31. T.Daniel, J.Xuan. Digit. Chem. Eng., 12, 100168 (2024); https://doi.org/10.1016/j.dche.2024.100168

32. Y.Yuan, D.Chaffart, T.Wu, J.Zhu. Engineering, 39, 45 (2024); https://doi.org/10.1016/j.eng.2023.11.024

33. E.F.Ruff, J.M.O.Zemke. J. Chem. Educ., 102 (4), 1457 (2025); https://doi.org/10.1021/acs.jchemed.4c01038

34. M.Bortz, K.Dadhe, A.Mitsos. Chem. Ing. Tech., 93, 1875 (2021); https://doi.org/10.1002/cite.202171202

35. A.Toniato, O.Schilter, T.Laino. Chimia, 77, 144 (2023); https://doi.org/10.2533/chimia.2023.144

36. C.Linchevski. Chem. Ing. Tech., 93, 2040 (2021); https://doi.org/10.1002/cite.202100099

37. OpenAI. New AI classifier for indicating AI-written text. 2023. https://openai.com/index/new-ai-classifier-for-indicating-ai-written-text/ (accessed 21 August 2026)

38. D.Weber-Wulff, A.Anohina-Naumeca, S.Bjelobaba, T.Foltýnek, J.Guerrero-Dib, O.Popoola, P.Šigut, L.Waddington. Int. J. Educ. Integr., 19 (1), 26 (2023); https://doi.org/10.1007/s40979-023-00146-z

39. V.S.Sadasivan, A.Kumar, S.Balasubramanian, W.Wang, S.Feizi. Can AI-Generated Text be Reliably Detected? arXiv, 2303.11156 (2023); https://arxiv.org/abs/2303.11156 (accessed 21 August 2026)

40. Turnitin. Understanding false positives within our AI writing detection capabilities. 2023; https://www.turnitin.com/blog/understanding-false-positives-within-our-ai-writing-detection-capabilities (accessed 21 August 2026)

41. Turnitin. Using the AI Writing Report; https://guides.turnitin.com/hc/en-us/articles/22774058814093-Using-the-AI-Writing-Report (accessed 21 August 2026)

42. Y.He, Y.Bu. Proc. Natl. Acad. Sci. USA, 123 (9), e2526734123 (2026); https://doi.org/10.1073/pnas.2526734123