Using AI in scientific articles
AI may be used as a supporting tool, but only transparently and responsibly. Artificial intelligence cannot be the author of an article, and all responsibility for the text always remains with a human.
In detail
Modern journals increasingly encounter the use of AI tools in manuscript preparation. The basic approach is this: AI may be used, but it should not be hidden if the tool actually participated in creating a significant part of the text, analysis, or visual content.
What is usually acceptable
- for language editing;
- for improving text structure;
- for drafting paragraphs;
- for assistance in technical analysis, if disclosed and verified by the author.
What is considered problematic
- listing AI as an author;
- hiding the use of AI where disclosure is required;
- including unverified facts, references, or conclusions;
- sending a confidential manuscript to services that may violate confidentiality rules.
COPE and WAME explicitly state that AI cannot be regarded as an author and cannot bear responsibility for a scholarly publication. Responsibility belongs only to real human authors.
What the author should do
- check the rules of the specific journal;
- if necessary, disclose this in Methods, Disclosure, Acknowledgements, or an Author Note;
- carefully re-check all wording, facts, and references.
What is important to remember
AI is a tool, not a co-author. It may be used only with transparent disclosure and full verification of the result.