The Scholarship of Research (SOR) recognises that generative artificial intelligence, Large Language Models (LLMs), machine-learning systems, and other AI-assisted technologies are increasingly used across the research lifecycle. These technologies may support legitimate scholarly activities, but their use must remain transparent, methodologically justified, ethically responsible, and subject to meaningful human oversight.

Because SOR specifically examines the methods, practices, ethical implications, and epistemological foundations of research, the Journal distinguishes between AI used to assist manuscript preparation and AI used as part of the research methodology, analysis, interpretation, or object of scientific investigation.

Human Responsibility and Accountability

Human authors remain fully responsible for the originality, accuracy, validity, integrity, interpretation, and reliability of all submitted work, regardless of whether artificial intelligence or AI-assisted technologies were used during the research or manuscript-preparation process.

AI-generated or AI-assisted output must never be treated as inherently accurate, authoritative, or independently verified. Authors are responsible for critically evaluating and validating all outputs before they are incorporated into scholarly work.

Responsibility for scientific reasoning, methodological decisions, interpretation, ethical compliance, and final conclusions must remain with human researchers.

Authorship

Artificial intelligence systems, Large Language Models, chatbots, machine-learning systems, and other automated technologies cannot be listed as authors or co-authors.

Authorship requires responsibilities that cannot be assigned to an AI system, including:

• approval of the final manuscript;
• accountability for the accuracy and integrity of the work;
• responsibility for methodological and interpretative decisions;
• management and disclosure of conflicts of interest;
• compliance with ethical and legal requirements;
• response to questions concerning the validity, provenance, or reliability of the published work.

AI systems must not be credited through the author byline or treated as responsible intellectual agents.

AI-Assisted Manuscript Preparation

Authors may use generative AI or AI-assisted technologies to support aspects of manuscript preparation, provided that such use does not replace human scientific judgment or responsibility.

Permitted uses may include, for example, language refinement, translation, structural assistance, summarisation undertaken under human supervision, or other forms of writing support.

Authors remain fully responsible for all generated or modified content.

Generative AI must not be used to:

• fabricate or falsify data, findings, analyses, references, quotations, or evidence;
• create fictitious ethical approvals, registrations, participants, datasets, or research procedures;
• misrepresent the provenance of scientific content;
• generate unsupported conclusions presented as established evidence;
• conceal methodological decisions or replace required human scientific interpretation;
• produce references that have not been independently verified by the authors.

Disclosure of AI Use

The use of generative AI or AI-assisted technologies to generate, substantially revise, translate, analyse, interpret, classify, code, synthesise, or otherwise materially contribute to intellectual or scientific content must be disclosed transparently.

Where AI is used primarily during manuscript preparation, the disclosure should identify:

• the name of the system or tool;
• the model or version, where available;
• the purpose for which it was used;
• the sections or aspects of the manuscript affected;
• the steps taken by the authors to review, verify, and take responsibility for the output.

A suitable disclosure statement is:

“During the preparation of this manuscript, the authors used [name and version of tool] for [purpose]. All AI-generated or AI-assisted content was reviewed and verified by the authors, who accept full responsibility for the final content of the manuscript.”

The use of conventional spelling, grammar-checking, formatting, reference-management, or similar tools that do not generate or substantially modify intellectual content does not normally require disclosure.

AI as Part of the Research Methodology

When artificial intelligence, machine learning, Large Language Models, Natural Language Processing, automated classification, generative models, multimodal systems, or related technologies form part of the research methodology, their use must be reported with sufficient detail to permit scientific evaluation and, where reasonably possible, reproducibility.

Depending on the nature of the study, authors should report:

• the name of the model, system, software, or platform;
• the model or software version;
• the provider or developer;
• the date or period of access when relevant to systems that may change over time;
• the scientific purpose for which the technology was used;
• the input data, corpus, prompts, or materials supplied to the system where scientifically relevant;
• relevant prompts, prompt architecture, system instructions, or interaction procedures where necessary for reproducibility;
• parameters, settings, thresholds, temperature, seeds, or other computational configuration where applicable;
• pre-processing and post-processing procedures;
• validation and quality-control procedures;
• human review, adjudication, or oversight mechanisms;
• methods used to evaluate accuracy, reliability, robustness, bias, or reproducibility;
• known limitations and sources of uncertainty.

Authors should provide sufficient methodological information to allow readers to understand how the AI system contributed to the production of the reported evidence or interpretation.

Human–AI Interpretation and Scientific Judgment

Where AI contributes to interpretation rather than merely computation, authors must explain how human and automated contributions were distinguished and integrated.

This requirement is particularly important for qualitative analysis, textual analysis, hypothesis generation, conceptual synthesis, evidence interpretation, and other research activities in which meaning or scientific judgment is generated rather than mechanically calculated.

Authors should describe, where relevant:

• which components of the interpretation were produced by human researchers;
• which components involved AI-assisted or automated processes;
• how disagreements between human and AI-generated interpretations were resolved;
• how reflexivity, contextual understanding, and domain expertise were preserved;
• how meaningful human oversight was maintained.

The use of AI does not remove the need for researcher judgment, methodological reflexivity, or scientific accountability.

Validation and Reproducibility

AI-generated outputs used as research data, classifications, predictions, interpretations, annotations, or analytical results must be validated using methods appropriate to the scientific question.

Validation may include, where appropriate:

• comparison with expert human assessment;
• independent replication;
• benchmark datasets;
• inter-rater or inter-method comparisons;
• sensitivity analyses;
• external validation;
• robustness testing;
• error analysis;
• assessment of bias, fairness, calibration, or uncertainty.

Authors should not assume that apparently coherent or plausible AI-generated output constitutes scientifically valid evidence.

Where exact reproducibility is limited because of proprietary models, stochastic outputs, model updates, or inaccessible system architecture, these limitations must be reported explicitly.

Prompts, Code, and Research Materials

Where prompts, analytical scripts, code, model configurations, or AI interaction protocols materially affect the results, authors should make them available whenever legally, ethically, and technically possible.

These materials may be included in the manuscript, supplementary files, or a recognised repository with a persistent identifier.

Authors must not disclose confidential, copyrighted, personally identifiable, or otherwise protected information merely for the purpose of methodological transparency. Where complete disclosure is not possible, the reason should be stated.

Images, Figures, Data, and Scientific Content

Generative AI or image-processing systems must not be used to fabricate, introduce, remove, obscure, or materially alter scientifically relevant information in images, figures, datasets, or other research materials.

AI-generated illustrative material must never be presented in a way that could reasonably cause readers to interpret it as original scientific evidence.

Where AI-assisted image processing is scientifically justified, its use must be disclosed in the Methods section and, where appropriate, in the relevant figure legend.

Authors must distinguish between:

• scientific images derived from original data;
• AI-assisted processing of original scientific images;
• synthetic or simulated scientific data;
• illustrative or conceptual AI-generated graphics.

Synthetic or simulated data must be clearly identified as such and must not be presented as observations obtained from real participants, experiments, or datasets.

AI-Generated or Synthetic Data

Where synthetic data generated through artificial intelligence or computational modelling form part of a study, authors must clearly identify their synthetic nature and explain the generation process, assumptions, validation procedures, and intended analytical role.

Synthetic data must not be represented as equivalent to empirical observations without appropriate scientific justification.

Authors should discuss relevant risks, including data leakage, memorisation of training data, unrealistic distributions, hidden bias, and limitations in external validity.

References and Evidence Verification

Authors are responsible for independently verifying every reference included in a manuscript.

AI-generated or AI-suggested references must not be cited unless the authors have confirmed:

• that the source genuinely exists;
• that the bibliographic information is accurate;
• that the cited source supports the claim for which it is used;
• that the source is appropriate and relevant.

Fabricated, unverifiable, or misleading references generated through AI systems may constitute a breach of research integrity.

Confidentiality, Privacy, and Intellectual Property

Authors must not upload confidential research information, identifiable participant data, unpublished third-party material, copyrighted content, proprietary information, peer-review material, or other restricted content to external AI systems unless appropriate legal authority, permissions, security safeguards, and data-protection arrangements are in place.

Particular care is required when external AI systems retain prompts, inputs, uploaded files, or user interactions for system improvement or other secondary purposes.

Authors remain responsible for compliance with applicable privacy, confidentiality, informed-consent, data-governance, copyright, contractual, and intellectual-property requirements.

Bias, Fairness, and Epistemic Responsibility

When AI systems contribute materially to research, authors should consider whether model design, training data, algorithmic architecture, or automated decision-making may introduce systematic bias or affect particular populations, languages, cultures, disciplines, or forms of knowledge.

Where relevant, manuscripts should address:

• algorithmic bias and fairness;
• representativeness of training or evaluation data;
• epistemic bias;
• limitations affecting underrepresented groups or languages;
• interpretability and explainability;
• accountability for automated outputs;
• the implications of delegating scientific judgment to computational systems.

Use of AI by Reviewers

Manuscripts submitted to The Scholarship of Research are confidential scholarly documents.

Reviewers must not upload manuscripts, figures, supplementary files, unpublished datasets, peer-review correspondence, or substantial manuscript content to publicly accessible or external generative AI systems when doing so could compromise confidentiality, copyright, intellectual property, privacy, or research integrity.

Reviewers must not delegate their scientific assessment to an AI system.

Peer-review reports must reflect the reviewer's own expert evaluation, reasoning, and judgment. Generative AI must not be used to produce a review that the reviewer has not independently formulated and verified.

Reviewers remain fully responsible for the accuracy, fairness, originality, and intellectual content of their review reports.

Use of AI by Editors

Editors must protect the confidentiality of submitted manuscripts and must not upload unpublished submissions, reviewer reports, supplementary material, or confidential editorial correspondence to external AI systems where confidentiality or data protection cannot be assured.

Artificial intelligence may not replace independent editorial judgment.

Editorial decisions concerning suitability, peer review, revision, acceptance, rejection, correction, or retraction must remain under human editorial responsibility.

Where AI-assisted tools are used for administrative, technical, similarity-detection, integrity-screening, or workflow-support purposes, their outputs must be interpreted by qualified human Editors and must not automatically determine editorial decisions.

AI Detection Tools

The Journal does not regard automated AI-content detection scores, standing alone, as definitive evidence that generative AI has been used improperly.

Where concerns arise regarding undisclosed or inappropriate AI use, Editors will consider the totality of available evidence and may request clarification, source materials, methodological documentation, version histories, code, prompts, or other information reasonably necessary to assess the concern.

Investigation of Undisclosed or Inappropriate AI Use

Suspected undisclosed, misleading, unethical, or scientifically inappropriate use of artificial intelligence or AI-assisted technologies will be evaluated under the Journal's Research and Publication Misconduct Policy and relevant COPE principles.

Depending on the circumstances, the Journal may:

• request clarification or additional disclosure;
• request prompts, code, underlying data, analytical records, or original research materials;
• require revision of the manuscript;
• reject the manuscript;
• publish a correction;
• publish an expression of concern;
• retract a published article;
• contact the authors' institution, ethics committee, funding organisation, or other relevant body where appropriate.

The seriousness of the response will depend on the nature of the AI use, the degree of transparency, its effect on the scientific record, and whether the integrity or reliability of the work has been compromised.

Principle of Meaningful Human Oversight

The Scholarship of Research recognises the scientific value of responsible innovation in artificial intelligence while maintaining that accountability for research must remain human.

AI systems may support scientific inquiry, but they must not obscure who made consequential methodological decisions, how evidence was interpreted, how outputs were validated, or who is responsible for the resulting scientific claims.

The Journal therefore requires meaningful human oversight whenever artificial intelligence materially contributes to research design, data generation, analysis, interpretation, reporting, or publication.

For questions concerning the use of artificial intelligence in research or manuscript preparation, please contact the Editorial Office of The Scholarship of Research at sor@ieditore.com.

Policy last updated: September 15, 2026.