Ethical AI Use in Dissertation Writing: A Practical Guide and Checklist

Using AI tools in research and composition is increasingly common. This guide explains the ethical use of AI in dissertation writing, showing how to balance academic integrity, transparency, and reproducibility while avoiding common pitfalls.

Practical guide to the ethical use of AI in dissertation writing

The primary concern for many graduate students and supervisors is how to apply the ethical use of AI in dissertation writing without compromising academic integrity. This section defines key terms, clarifies responsibilities, and outlines where institutional policies and publication ethics intersect.

Key definitions and entities

Terms to understand include: AI-assisted writing (use of large language models or other generative tools to draft or edit text), provenance (tracking where outputs and data come from), attribution (declaring the role of tools), and reproducibility (ability for others to recreate methods). Relevant organizations include institutional review boards (IRBs), the Committee on Publication Ethics (COPE), and discipline-specific bodies such as the American Psychological Association (APA) or IEEE for technical fields.

CLARITY Checklist: a named framework for responsible AI use

Apply the CLARITY Checklist before, during, and after using AI tools in research writing:

• Cite inputs and outputs: Record prompts, model versions, and sources used to generate content.
• Limit scope: Use AI for support tasks (summaries, formatting, code scaffolding) rather than for original arguments or data interpretation without verification.
• Acknowledge use: Declare AI assistance in methodology or acknowledgments per institutional guidance.
• Review and validate: Independently verify facts, references, analysis, and code produced by AI.
• Inspect for bias and privacy risks: Evaluate outputs for demographic bias and ensure no private or sensitive data was exposed.
• Track versions: Keep versioned records of prompts, outputs, edits, and collaborator input.
• Yield approval: Discuss AI use with supervisors and seek formal sign-off where required.

How this aligns with best practices

Many journals and institutions now expect transparency about analytical tools. For guidance on publication ethics and disclosure norms, see the Committee on Publication Ethics (COPE) for best-practice statements and resources: Committee on Publication Ethics (COPE).

Practical steps to implement ethical AI use in a dissertation workflow

Step-by-step actions

1. Map where AI might be used (literature search, drafting, editing, coding) and document intended use cases.
2. Create a prompt and output log: save prompts, timestamps, model identifiers, and outputs in a reproducible folder or lab notebook.
3. Verify: cross-check facts, citations, and analyses with primary sources and independent methods.
4. Declare AI contributions in the thesis methods or acknowledgments and in any manuscripts submitted for publication.
5. Consult supervisors and institutional policies; submit documentation to an IRB if human subjects or sensitive data are involved.

Real-world scenario

A doctoral student used a language model to generate a first draft of a literature review. The student kept the original prompts and outputs, verified each citation against primary sources, replaced any incorrect or hallucinated statements, and added a methods subsection describing the AI assistance and the verification process. The supervisor approved the approach, and the final dissertation included a short declaration explaining the role of generative tools.

Common mistakes and trade-offs when using AI tools

Common mistakes

• Failing to verify AI-generated references or claims, leading to fabricated citations.
• Using AI to write original analysis without transparency or supervisor approval.
• Including sensitive data in prompts that expose participant information or violate consent agreements.

Trade-offs to consider

Using AI can speed drafting and improve clarity, but it creates tasks for verification and documentation. Relying heavily on AI for conceptual work may undermine training opportunities for critical thinking. Decisions should weigh time savings against the need for rigorous validation and the educational objectives of the dissertation.

Practical tips for students and supervisors

• Keep a searchable prompt-and-output log stored with research files to support reproducibility and review.
• Use AI for iterative, non-original tasks (formatting, grammar, summarizing long texts) and avoid outsourcing interpretation of data without independent checks.
• Include a short declaration in the dissertation that explains what was generated by AI and what was produced by the researcher.
• Discuss AI use early with supervisors and, where relevant, with the IRB or ethics committee to prevent compliance issues.

Core cluster questions

• How should AI assistance be documented in academic theses?
• What policies do universities commonly require for AI use in research?
• How to verify AI-generated references and data analyses?
• How does AI use affect authorship and contribution statements?
• What privacy and consent issues arise when using AI with sensitive datasets?

FAQ

What is the ethical use of AI in dissertation writing?

Ethical use means being transparent about AI assistance, verifying and citing outputs, protecting participant privacy, and following institutional and disciplinary guidance on authorship and disclosure.

Should AI-generated text be cited or acknowledged in a dissertation?

Yes. A clear statement in the methods or acknowledgments is recommended describing the tool, version, and the nature of the assistance. Keep logs of prompts and outputs for reproducibility and review.

Can AI replace supervisor guidance or original analysis?

No. AI should not replace supervisory mentorship or original scholarly analysis. Use AI as a tool for drafting, organization, or troubleshooting, while ensuring that intellectual contributions and critical reasoning remain the author’s responsibility.

How to verify references and results produced by AI tools?

Verify every citation and factual claim against primary sources. Re-run analyses independently when AI is used for code or statistical suggestions, and document oversight steps in research records.

How to handle privacy and data protection when using AI in research?

Avoid inputting identifiable or sensitive participant data into third-party models unless the platform and consent permits it. Consult an IRB and institutional data governance policies before using AI with human-subjects data.