Generative AI Policies

POLICY ON THE ETHICAL USE OF GENERATIVE ARTIFICIAL INTELLIGENCE

1. Context and Foundational Framework

This policy aligns with established global publication ethics frameworks, including STM recommendations on AI classification in academic manuscripts, Elsevier's directives on AI integration within peer review, and the World Association of Medical Editors (WAME) guidance on chatbots and scholarly publishing.

Jurnal AGRO acknowledges the utility of generative AI platforms (e.g., ChatGPT, Claude, Gemini, Copilot) in streamlining research workflows, assisting data structuring, and refining linguistic clarity—particularly for non-native English scholars. However, because generative AI operates via probabilistic modeling rather than factual reasoning, it introduces critical vulnerabilities:

  • Algorithmic Bias and Hallucination: Generation of unverified facts, fabricated references, or undetected systemic biases.

  • Attribution Deficits: Inconsistent or invalid sourcing, quotation, and academic attribution.

  • Data Security and Intellectual Property Risks: Exposure of proprietary findings to third-party tools that lack robust data protection guarantees or that repurpose user inputs for model training without consent.

Accordingly, Jurnal AGRO enforces clear operational boundaries to safeguard scholarly rigor and ethical integrity.

2. Directives for Authors

  • Authorship and Accountability: Generative AI tools cannot be credited as authors or co-authors. Authorship confers legal, ethical, and scientific accountability—obligations that only human researchers can assume.

  • Permissible Applications: AI assistance is strictly confined to preliminary conceptual exploration, linguistic and stylistic refinement, literature categorization, and coding support. All AI outputs must undergo comprehensive human verification.

  • Prohibited Practices: Submitting unedited AI-generated prose or code, substituting empirical findings with unverified synthetic datasets, and generating or altering visual assets (including charts, tables, medical images, formulas, and raw datasets) via generative AI are strictly prohibited.

  • Mandatory Disclosure: Authors must explicitly declare any AI usage in the Methods or Acknowledgements section, specifying the tool name, version number, operational scope, and research purpose.

3. Directives for Editors and Peer Reviewers

  • Confidentiality and Data Protection: Editors and peer reviewers are strictly prohibited from uploading unpublished manuscripts, supplementary files, images, or proprietary data into public or third-party AI platforms. Doing so constitutes a direct breach of manuscript confidentiality and authorial intellectual property rights.

  • Critical Evaluation: Manuscript evaluation, substantive critique, and editorial decision-making must remain exclusively human-driven. Reviewers may only utilize AI tools to polish the language and readability of their written feedback, retaining full responsibility for the content and judgment of their reports.