Using AI Responsibly in Your Literature Review
AI assistants are now part of how many students read, search and write. They can save time in a literature review, but they can also introduce errors that are hard to spot and harder to explain to a supervisor or examiner. In this blog, we'll look at where AI tools genuinely help, where they fall short, and how to use them while keeping your work honest and your own.
Check the Rules Before You Start:
- Read your university's policy on AI use, along with any guidance from your department or supervisor. Policies differ widely, and some require you to declare AI use in your thesis.
- If you plan to submit to a journal, check the publisher's author guidelines too. Many journals now say how AI may be used and how it should be disclosed.
- When the policy is unclear, ask your supervisor before you rely on a tool, not after.
Use AI to Explore, Not to Cite:
- AI tools can help you brainstorm search keywords, find related terms and synonyms, and map out the sub-themes of an unfamiliar topic.
- Treat these suggestions as leads, not as sources. Every paper you cite must be one you have found in a database such as Google Scholar, Scopus or your library catalogue and have read yourself.
- Never copy a reference from an AI response straight into your reference list. AI tools can produce citations that look real but do not exist, or attach the wrong authors, year or journal to a real paper.
Verify Every Claim Against the Original Source:
- If an AI tool summarises a paper, open the paper and check the summary against it. Look closely at the sample, method, findings and limitations, because these are the details most often misstated.
- Be careful with numbers. Sample sizes, percentages and effect sizes should always come from the original article, never from an AI summary.
- A useful habit is to note, for each source, what you checked personally. This makes it easy to answer your supervisor's questions with confidence.
Keep the Thinking and Writing Your Own:
- The value of a literature review lies in your synthesis: how you group studies, where you see agreement and conflict, and what gap you identify. That judgement is what examiners assess, and it should come from you.
- You can ask an AI tool to point out unclear sentences or suggest a better structure for a paragraph. Rewriting whole sections with AI and submitting them as your own work is a different matter, and it may breach your institution's rules.
- Write your first draft yourself. Use AI afterwards, if permitted, as a second reader, not as the author.
Protect Data and Unpublished Work:
- Do not paste confidential material into public AI tools. This includes interview transcripts, survey responses, unpublished manuscripts you are reviewing, and your supervisor's unpublished ideas.
- If your research involves human participants, check whether your ethics approval allows participant data to be processed by external tools.
Record and Declare Your AI Use:
- Keep a simple log of which tool you used, when, and for what purpose, for example "keyword brainstorming for Chapter 2". This takes a few minutes and protects you if questions arise later.
- Declare AI use as your institution or journal requires, usually in the methods section, acknowledgements or a separate declaration.
AI tools can make a literature review faster, but they cannot take responsibility for its accuracy. That responsibility stays with the researcher. By following your institution's rules, verifying every source and claim, keeping the synthesis your own, protecting sensitive data and recording your AI use, you can benefit from these tools without putting your academic integrity at risk.