Artificial intelligence policy

Artificial Intelligence and Use of Large Language Models (LLM)

This policy regulates the use of artificial intelligence (AI) tools—including large language models (LLMs), chatbots, and image generators—at all stages associated with publishing in Revista ALP: research, manuscript drafting, submission, peer review, and the editorial process.

1.1 Guiding Principles (In alignment with Universidad de Costa Rica's guidelines)

  • Human Responsibility: Intellectual, ethical, and academic responsibility for submitted and published content rests exclusively with the authors. The use of AI does not substitute for professional and academic judgment, nor does it exempt individuals from responsibility.
  • Transparency and Traceability: Any use of AI that contributes to the preparation of the manuscript (text, analysis, translation, images, tables, code, etc.) must be declared in a clear, sufficient, and verifiable manner.
  • Integrity of the Scientific Record: The use of AI to fabricate, falsify, alter, or manipulate data, results, evidence, citations, references, images, or any other element in a misleading way is rejected.
  • Confidentiality and Content Protection: Materials submitted to Revista ALP and the peer review process are confidential. Manuscripts (or parts thereof) must not be shared with AI tools that compromise this confidentiality, especially through free services or those lacking institutional guarantees.
  • Prohibition of Editorial Process Manipulation via “Hidden Prompts”: It is strictly prohibited to insert hidden instructions or mechanisms intended to manipulate editorial judgment or review outcomes.

 

1.2 Authorship, Citation, and Responsible Use of AI

  1. A) AI and Authorship
  • AI systems do not meet the authorship criteria of Revista ALP and, therefore, cannot be listed as authors or co-authors (consistent with the Declaración de Heredia, Penabad et al., 2024).
  • Revista ALP does not accept AI-generated content presented as academic "sources" in the reference list.
  1. B) AI and Citation
  • Authors must only cite verifiable academic sources (books, articles, official documents, data, etc.).
  • The use of AI as a tool must be declared (see 1.3). Bibliographic references such as “ChatGPT/AI as a source of knowledge” are not permitted.
  1. C) Responsibility for Accuracy and Originality
  • The use of AI does not exempt authors from their ethical, scientific, and academic responsibility regarding the truthfulness and precision of statements and data, the originality of the text (including paraphrasing), the integrity of results, correct attribution and citation, and compliance with copyrights and tool terms of use.

 

1.3 Declaration of Use of Artificial Intelligence

The AI Use Declaration is mandatory whenever generative AI tools or equivalent systems are used for any component of the manuscript (text, analysis, translation, tables, images, code, etc.).

Where to include it:

  • In the manuscript: In the ‘Methods’ section when AI is part of the design, literature review, analysis, corpus construction, coding, modeling, or data processing, etc.
  • Explicit Section: Also in a dedicated section titled “AI Use Declaration” at the end of the manuscript. If used only for writing support, copyediting, or translation, it must also be indicated here.

Minimum mandatory information to include in the ‘Methods’ section:

  1. Tool used: Name of the tool and provider.
  2. Model and version (or available equivalent) and date(s) of use.
  3. Purpose and scope of use, distinguishing between: (i) stylistic/formal use (e.g., linguistic correction, rewriting, translation, clarity improvement) and/or (ii) substantive use (e.g., generating drafts of sections, literature review, data modeling/processing, data analysis/coding, creation of tables or figures, etc.).
  4. Inputs provided to the AI (e.g., original text fragments, anonymized data, code) and measures taken to protect sensitive or confidential information.
  5. Human verification and control measures: How accuracy was checked, how errors/biases were corrected, how citations were verified, and how methodological consistency was ensured.

Recommended Best Practices:

  • Maintain a record of prompts, parameters, and relevant outputs for auditing or replicability purposes.
  • Be prepared to provide this information to the journal if requested during editorial evaluation (respecting confidentiality and applicable regulations).

Suggested Declaration Model (Authors): “During the preparation of this manuscript, [Tool] ([Provider], model [X], version [Y], date[s] of use [Z]) was used for [specific purpose]. Its use was limited to [sections/tasks]. As an author/co-author, I declare that I have fully verified and edited the generated results, checked the accuracy of the information and references, and assume full responsibility for the final content.” [Indicate, if applicable, that no personal/sensitive data was entered or how it was protected].

 

1.4 Use of AI by Reviewers

  • Non-delegation of judgment: Reviewers cannot delegate the critical evaluation of a manuscript to an AI. The verdict is a human, personal, and non-transferable responsibility.
  • Permitted use: Auxiliary use is permitted only to improve the clarity of the review report's writing (e.g., copyediting the verdict itself), provided it does not replace expert judgment, does not generate the substantive evaluation, and maintains absolute manuscript confidentiality.
  • Prohibition on uploading manuscripts to AI: It is strictly forbidden to input the manuscript, parts of the manuscript, tables, figures, or unpublished data into free AI systems or those without institutional guarantees that could jeopardize the confidentiality and originality of the article.
  • Mandatory declaration by reviewers: If AI was used (even in an auxiliary manner), it must be declared in the report to the responsible editor, including: tool, model and version, purpose, verification measures, and explicit confirmation that the manuscript was not uploaded to free or non-institutional AI systems.

Suggested Declaration Model (Reviewers): “To support the drafting of this report, I used [Tool] (model [X], version [Y]) solely for [auxiliary purpose]. Neither the manuscript nor confidential information was uploaded to the tool. The critical analysis and the conclusions of this report are entirely my responsibility as a reviewer.”

1.5 Use of AI by Revista ALP and its Editorial Team

  • Revista ALP may use AI tools for editorial support (e.g., administrative tasks, editorial communications, internal copyediting support, metadata organization, similarity checking, among others).
  • All editorial decisions (acceptance, rejection, requests for revision, selection of reviewers, etc.) will be made exclusively by humans and not by automated systems.
  • The journal declares the use of tools such as Microsoft 365 Copilot and ChatGPT under institutional or paid licensing terms, ensuring data protection safeguards and content confidentiality.
  • Any internal use of AI will be conducted following data minimization criteria, avoiding the exposure of manuscripts to services that could compromise confidentiality or originality.

 

1.6 Non-compliance, Editorial Evaluation, and Measures

Failure to provide the AI Use Declaration when required, the misleading use of AI (including manipulation via “hidden prompts”), or any practice that compromises academic integrity may be considered a serious breach of editorial standards.

Upon suspicion or evidence of such practices, Revista ALP may:

  1. Request clarifications and evidence of verification (e.g., detailed explanation of use, general prompts, logs, or previous versions).
  2. Require corrections.
  3. Reject the manuscript.
  4. If already published, apply editorial measures (corrections, editorial notes, retractions, etc.) in accordance with the Revista ALP Code of Ethics, Good Practices, and Transparency, and COPE guidelines.