Generative AI Policy

INTRODUCTION

This policy is based on and refers to the guidelines outlined in the Generative AI Policies for Journals, as provided by:

Jurnal Fisika Unand (JFU) understands the importance of artificial intelligence (AI) and its potential to help authors with their research and writing processes. JFU is excited about the new possibilities that generative AI tools bring, especially for generating ideas, speeding up research, analyzing results, improving writing, organizing submissions, assisting authors who write in a second language, and accelerating the research and sharing process. JFU is offering guidance to authors, editors, and reviewers on the use of such tools, which may evolve as the AI field continues to develop rapidly.

Generative Artificial Intelligence (AI) tools, including large language models (LLMs) and multimodal models, are continually developing and evolving, particularly in their applications for businesses and consumers. While generative AI has significant potential to enhance authors' creativity, it is important to acknowledge the risks associated with the current generation of these tools. Generative AI can produce a wide variety of content, including text generation, image synthesis, audio generation, and synthetic data. Notable examples of such tools include ChatGPT, Copilot, Gemini, Claude, NovelAI, Jasper AI, DALL-E, Midjourney, and Runway.

Some of the risks associated with the operation of generative AI tools today are:

  • Inaccuracy and bias: Generative AI tools are fundamentally statistical in nature rather than factual. Consequently, they can introduce inaccuracies, falsehoods (often referred to as hallucinations), or biases that may be difficult to detect, verify, and rectify.

  • Lack of attribution: Generative AI frequently fails to adhere to established practices in the global scholarly community for accurate and precise attribution of ideas, quotations, or citations.

  • Confidentiality and intellectual property risks: Currently, generative AI tools are often employed on third-party platforms that may not provide adequate standards for confidentiality, data security, or copyright protection.

  • Unintended uses: Providers of generative AI may repurpose input or output data generated from user interactions (for instance, for AI training). This practice has the potential to infringe upon the rights of authors and publishers, among others.

AUTHORS

Authors may use generative AI tools (e.g., ChatGPT, GPT models) for specific tasks, such as enhancing the grammar, language, and readability of their manuscripts. However, authors remain responsible for the originality, validity, and integrity of their submissions. When using generative AI tools, it is essential that authors do so responsibly and adhere to our journal's editorial policies on authorship and publication ethics. This responsibility encompasses reviewing the outputs produced by any AI tools and ensuring the accuracy of the content.

JFU endorses the responsible use of generative AI tools, ensuring that high standards of data security, confidentiality, and copyright protection are maintained in instances such as the following:

  • Idea generation and idea exploration

  • Language improvement and readability refinement

  • Interactive online search with LLM-enhanced search engines

  • Literature classification

  • Coding assistance (e.g., Python, MATLAB for physics simulations and data processing)

Authors are responsible for ensuring that the content of their submissions meets the required standards of rigorous scientific and scholarly assessment, research, and validation, and is created by the author.

Generative AI tools should not be credited as authors, as they cannot assume responsibility for the content submitted or manage copyright and licensing agreements. Authorship entails accountability for the content, consent to publication under a publishing agreement, and contractual assurances regarding the integrity of the work, among other essential principles. These uniquely human responsibilities cannot be fulfilled by generative AI tools.

Authors must clearly acknowledge any use of generative AI tools in their articles by including a statement that specifies the full name of the tool (along with its version number), how it was used, and the reason behind it. For article submissions, this statement should be placed in a dedicated section titled "Declaration of Generative AI and AI-Assisted Technologies in the Writing Process" prior to the References section, or within the Methods or Acknowledgments section. This transparency allows editors to assess the employment and responsible use of generative AI tools. JFU will maintain discretion over the publication of the work to ensure that integrity and guidelines are upheld.

If an author intends to use an AI tool, they must ensure that it is suitable and robust for their intended purpose. Additionally, they should verify that the terms associated with such a tool offer adequate safeguards and protections, particularly concerning intellectual property rights, confidentiality, and security.

Authors should avoid submitting manuscripts that use generative AI tools in ways that compromise fundamental researcher and author responsibilities, for example:

  • text or code generation without rigorous revision and verification

  • synthetic data generation to substitute missing primary data without a robust methodology

  • generation of any type of content that is inaccurate, including abstracts or supplemental materials

These types of cases may be subject to editorial investigation.

In accordance with Elsevier and Taylor & Francis policies, JFU currently prohibits the use of generative AI in the creation, alteration, or manipulation of primary images, figures, and research data. The term “images and figures” encompasses photos, experimental charts, characterization spectra, medical images, image snippets, computer code, and physical formulas. “Manipulation” refers to augmenting, concealing, moving, removing, or introducing specific features within an image or figure. The only exception is when generative AI is explicitly part of the research design or methodology (e.g., AI-based image-processing algorithms or machine-learning models in physics research), in which case it must be fully declared, detailed, and reproducible in the Methods section.

Human oversight and transparency must consistently inform the use of generative AI and AI-assisted technologies throughout every stage of the research process. Research ethics guidelines are continuously being revised to reflect advancements in generative AI technologies. JFU will continue to update our editorial guidelines as both the technology and ethical standards in research develop.

EDITORS AND PEER REVIEWERS

JFU is committed to maintaining the highest standards of editorial integrity and transparency. Editors and peer reviewers who use manuscripts in generative AI systems could risk compromising confidentiality, ownership rights, and privacy, including personal information. Consequently, editors and peer reviewers are prohibited from uploading files, images, or information from unpublished manuscripts into generative AI tools. Non-compliance with this policy may violate the intellectual property rights of the rightsholder.

Editors

Editors play a crucial role in maintaining the quality and integrity of research content. Consequently, it is imperative that editors maintain the confidentiality of submission and peer-review details.

The use of manuscripts in generative AI systems could pose significant risks to confidentiality, as well as potential infringements of proprietary rights, data security, and other concerns. Therefore, editors are prohibited from uploading unpublished manuscripts, along with any associated files, images, or information, into generative AI tools.

Peer Reviewers

Peer reviewers, who are selected as subject-matter experts, should refrain from utilizing generative AI to evaluate, synthesize, or condense submitted articles, or any portion of them, when writing their reviews. Consequently, peer reviewers must not upload unpublished manuscripts or project proposals, nor any associated files, images, or information, into generative AI tools.

Generative AI may be used only to enhance the language and readability of the review report itself; however, peer reviewers will always be accountable for ensuring the accuracy, objectivity, and integrity of their assessments.