Artificial intelligence in scientific research: a practical guide to tools and ethical considerations

Abstract

Artificial intelligence (AI), including generative AI and large language models (LLMs), is rapidly transforming scientific research. It is increasingly being integrated into multiple stages of the research process. The public release of ChatGPT in November 2022 contributed to increased adoption of these technologies, creating new opportunities for literature discovery, data analysis, and research workflow support. However, many researchers remain uncertain about which AI tools to use and how to use them ethically at each stage of the research process. This editorial review provides a practical, stepwise framework for integrating AI across the research lifecycle, from idea generation and literature searching through data analysis and scientific writing to publication, aligned with the traditional research steps. We discuss the applications of commonly used AI tools, including general-purpose LLMs, literature-search and citation-analysis tools, and AI-assisted writing and translation tools. AI-assisted workflows may improve research efficiency, facilitate evidence synthesis, support scientific writing, and support researchers conducting research in a non-native language. However, important limitations remain, including hallucinations and fabricated information or references, algorithmic bias, inadequate source verification, risks of plagiarism and inappropriate reuse of existing material, confidentiality and data-privacy risks, and potential overreliance on automated outputs. Current guidance from the International Committee of Medical Journal Editors (ICMJE), the Committee on Publication Ethics (COPE), and the World Association of Medical Editors (WAME) emphasizes that AI tools should not be credited as authors and that authors should disclose their use in accordance with applicable journal policies. AI should be regarded as an assistive technology rather than a substitute for human expertise, critical appraisal, scientific judgment, and accountability. Responsible integration requires appropriate tool selection, verification of AI-generated outputs, protection of sensitive information, transparent reporting consistent with applicable journal policies, and continuous human oversight. Clear institutional and journal policies are also needed to support responsible and ethical AI use in scientific research.