LLM IN SCIENCE: ATTRIBUTION AND REPRODUCIBILITY
DOI:
https://doi.org/10.5281/zenodo.17534593Keywords:
large language models, generative artificial intelligence, scientific research, authorship, transparency, reproducibility, academic integrity, AI ethicsAbstract
Large Language Models (LLMs) are transforming research and education by enabling new forms of text generation, analysis, and automation. This study reviews current policies and ethical standards regarding the use of generative AI in scientific research, focusing on authorship attribution, transparency, and reproducibility. It analyzes leading international journal policies (Science, Nature, Springer Nature, Elsevier) and funding agency guidelines (NIH, NSF), as well as the Ukrainian Ministry of Education and Science (2025) recommendations. The findings show a global consensus: AI cannot be an author, and its use must be disclosed. To ensure reproducibility, researchers are encouraged to provide datasets, code, model parameters, and prompts. Risks such as hallucinations, fake citations, and model drift require mitigation via provenance tracking, DOI verification, watermarking, and ethical disclosure. Adherence to COPE, ICMJE, NISO, FAIR/RO-Crate, ORCID, and ROR standards ensures accountability and integrity in AI-assisted research. The paper concludes that responsible integration of AI into science and education can enhance innovation while preserving trust and reproducibility.
References
Nature Editorial Board. (2023). Tools such as ChatGPT threaten transparent science; here are our ground rules for their use. Nature, 613(7945), 612. https://doi.org/10.1038/d41586-023-00191-1
Springer Nature. (2023). Statement on the use of generative artificial intelligence in manuscripts. Springer Nature. https://group.springernature.com/gp/group/media/press-releases/archive-2023/first-ai-generated-book/26189712
Elsevier. (2023). Principles for use of generative artificial intelligence tools in the publication process. https://www.elsevier.com/about/policies-and-standards/the-use-of-generative-ai-and-ai-assisted-technologies-in-writing-for-elsevier
COPE Council. (2023). Authorship and AI tools – COPE position statement. Eastleigh: Committee on Publication Ethics (COPE).; 2023.https://publicationethics.org/guidance/cope-position/authorship-and-ai-tools
International Committee Of Medical Journal Editors. (2024). Recommendations for the conduct, reporting, editing and publication of scholarly work in medical journals. Ewha medical journal, 47(4), e48. https://doi.org/10.12771/emj.2024.e48
National Institutes of Health (NIH), Office of Extramural Research. (2024). Guidance on the use of generative AI in peer review and grant applications.
National Science Foundation (NSF), Directorate for Technology, Innovation and Partnerships. (2024). Responsible use of generative artificial intelligence in research and education. https://www.nsf.gov/focus-areas/ai#:~:text=Share,Find%20funding%20in%20AI
Міністерство освіти і науки України. (2025). Щодо використання систем штучного інтелекту у закладах освіти: лист №1/779-23 24.04.2025. .https://osvita.ua/legislation/list/340/
Wilkinson, M. D., Sansone, S. A., Schultes, E., Doorn, P., Bonino da Silva Santos, L. O., & Dumontier, M. (2024). FAIR principles: Reproducibility and transparency in data and AI research. Sci Data. 11(1), 145.
Thorp, H. H. (2023). ChatGPT is fun, but not an author. Science. 379(6630), 313. https://doi.org/10.1126/science.adg7879
Resnik, D. B., Hosseini, M., & Rasmussen, L. M. Using AI to write scholarly publications. (2023). Accountability in Research, 31(7):1–9. https://doi.org/10.1080/08989621.2023.2168535
Міністерство освіти і науки України & Міністерство цифрової трансформації України. (2025). Рекомендації щодо відповідального впровадження та використання технологій штучного інтелекту в закладах вищої освіти https://www.kmu.gov.ua/news/mintsyfry-ta-mon-razom-z-ekspertamy-rozrobyly-rekomendatsii-shchodo-vidpovidalnoho-vykorystannia-shi-u-vuzakh
