Proposed Guidelines for the Responsible Use of AI in Research

Responsible and Informed Use

AI tools are increasingly integrated into research practice and are widely accessible. However, widespread accessibility does not imply deep understanding of underlying mechanisms. Given that many researchers use AI without fully understanding how it operates, research institutions are encouraged to promote informed and reflective use.

Researchers are expected to develop a basic understanding of the capabilities and limitations of AI systems, avoid uncritical reliance on outputs, and ensure that AI serves as a support tool rather than a substitute for disciplinary expertise.

Research institutions are encouraged to provide structured guidance to prevent fragmented or superficial practices and reduce uncertainty regarding data protection, intellectual property, privacy, and cost-related concerns.

Transparency and Mandatory Disclosure

Full transparency in AI use is required.

All researchers should explicitly declare whether AI tools were used, for which tasks (e.g., drafting, coding assistance, statistical support, translation), and to what extent AI contributed to the final output. The guiding principle is: use is permissible, but declaration is expected.

Where relevant, prompts and AI-assisted processes should be documented and, when appropriate, published as part of research data to support traceability and reproducibility.

Concealment of AI use, including hiding reliance or misrepresenting authorship, constitutes a breach of research integrity and should be addressed under existing misconduct frameworks.

Human Responsibility, Oversight, and Intellectual Integrity

Artificial Intelligence is a support tool in research and does not assume intellectual responsibility. Scholarly accountability, methodological judgment, and interpretation remain human obligations.

Researchers are responsible for the accuracy, validity, and integrity of their work, regardless of AI use. AI-generated outputs should be critically examined and verified, particularly in light of concerns about black box scenarios and reproducibility.

Over-reliance on AI may contribute to loss of expertise, reduced analytical engagement, and diminished human creativity. AI should therefore augment, not replace, disciplinary competence and conceptual development. Active cognitive engagement and intellectual authorship must remain with the researcher.

Research institutions are encouraged to reinforce these principles through clear accountability standards and appropriate monitoring of AI use and output quality.

Safeguarding Research Quality and Credibility

Efficiency gains should not compromise academic standards.

While AI can increase efficiency and save time, research institutions are encouraged to explicitly prioritize quality over quantity. Concerns about a potential shift from quality to quantity or speed and declining research credibility require proactive safeguards.

Researchers should use AI to enhance clarity (e.g., grammar, structure) without outsourcing intellectual substance. Productivity gains should not undermine rigor, originality, or methodological robustness.

Institutional evaluation systems are encouraged to reinforce standards of quality and credibility rather than incentivizing speed alone.

Reproducibility and Documentation

AI-assisted research processes should be thoroughly documented to address concerns related to reproducibility and opacity.

Researchers should record the tools used and their versions, document relevant prompts when they significantly shape results, and ensure that AI-supported analyses remain replicable where possible.

Research institutions may consider establishing repositories or standardized documentation formats for AI-supported research workflows to enhance transparency and shared learning.

Ethical Use and Prevention of Misconduct

Misuse of AI undermines academic trust and integrity.

Research institutions are encouraged to clearly define and address plagiarism facilitated by AI, undisclosed AI authorship, and misrepresentation of independent intellectual work. Clear distinctions should be maintained between legitimate assistance (e.g., grammar refinement, coding support) and unethical substitution of scholarly contribution.

Transparent communication and consistent enforcement mechanisms are necessary to prevent abusive practices.

Data Protection, Security, and Risk Awareness

AI use should comply with data protection and cybersecurity standards.

Given concerns regarding data protection, intellectual property, privacy, malware, and technological vulnerabilities, researchers should ensure that sensitive or confidential data are not uploaded to external AI systems without appropriate safeguards.

Research institutions are encouraged to provide secure, institutionally approved AI tools, offer guidance on safe data handling, and regularly review emerging technological risks, particularly as AI systems evolve rapidly.

Equity, Access, and Institutional Support

Unequal access to AI tools can create disparities in research opportunities.

Research institutions are encouraged to provide institutional access to approved AI systems, offer structured training programs, and establish dedicated AI advisors or experts within faculties or departments.

Workshops, training sessions, and ongoing professional development are essential to ensure that AI integration is informed, critical, and equitable.

Emotional and Cultural Dimensions of AI Integration

AI adoption affects researchers not only technically but also emotionally and culturally.

Responses may range from enthusiasm and excitement to anxiety, distrust, fear of falling behind, or feeling overwhelmed. Research institutions are encouraged to create spaces for open dialogue about these reactions and avoid normalizing either uncritical enthusiasm or blanket resistance.

Responsible AI governance includes fostering reflective academic cultures that support critical engagement.

Continuous Review and Adaptive Governance

AI technologies develop rapidly and may outpace institutional regulation.

Given that AI systems evolve rapidly and that lack of oversight was identified as a concern, research institutions are encouraged to implement mechanisms to monitor AI use and evaluate associated risks on an ongoing basis.

Ongoing evaluation should help ensure that AI integration remains aligned with research integrity, quality standards, and institutional responsibility.