AIMLS: standardized graphic labeling to disclose the use of artificial intelligence in scientific-academic manuscripts

Abstract

English translation of the Spanish-language preprint (original version): https://doi.org/10.5281/zenodo.23273378 Introduction: The use of artificial intelligence (AI) in scientific-academic writing continues to grow steadily, with and without editorial policies regulating it. A disclosure drafted without standardized vocabulary is not standardized, comparable, or itemized, and therefore does not produce a datum that can be counted. Objective: To develop an open-access tool that generates, based on what the author declares, a standardized graphic label accompanied by its textual disclosure in prose, to communicate the use of AI in the preparation of manuscripts in a clear, explicit, and visible way. Methods: We adopted a classification of eight declarable AI use activities according to the STM Association, and we added an optional four-grade ordinal scale of the extent of each use (minimal, partial, majority, and total) with respect to each task. We designed a monochrome label with fixed-rule alphanumeric notation, always generated in English, with the codes ordered according to the classification and a QR code that links to the system’s documentation. We developed the software, verified the consistency between the label and the prose through 1,337 automated tests, and illustrated its operation with an example case. Results: We obtained AIMLS (AI Manuscript Labelling System), an HTML application that runs in any browser, requires no installation or registration, and does not store or transmit the declared information. It generates a matching label and draft textual disclosure, whose notation allows the declared use of AI to be compared and counted. Its limitations include dependence on editorial adoption, visual accessibility, the voluntary and unverifiable nature of disclosures, and the need to update the taxonomy. Conclusion: The tool converts the disclosure of AI use into a visible, standardized, and portable label that complements, without replacing, editorial policies and prose disclosure; it constitutes a mechanism for communication, not certification or detection. As future work, we propose its integration into editorial platforms such as Open Journal Systems and scientific metadata standards, as well as its validation with users and periodic updating.