Using AI in Postgraduate Research Without Crossing the Integrity Line

The direct answer: the line sits at authorship of thought. AI may accelerate your searching, summarising, organising, and polishing; it may not do your thinking, your analysis, or your writing-as-argument for you. Use it as an instrument and disclose how — the same way you would any research instrument — and you are on solid ground at almost any university.

Masters and PhD students are caught between two bad messages: supervisors who treat every AI mention as misconduct, and peers who quietly paste whole chapters out of a chatbot. Both are wrong, and both are avoidable with a clear framework.

What is AI genuinely good for in research?

• Discovery: finding and triaging relevant literature far faster than manual database crawling.
• Comprehension: breaking down a dense 40-page paper so you can decide whether it deserves a full read.
• Organisation: structuring notes, mapping how sources relate, planning chapters.
• Language: tightening sentences you wrote, improving clarity and flow — the same service a human editor legitimately provides.

Where exactly is the line?

Three tests, applied honestly, settle nearly every case:

• The authorship test. If the intellectual move — the claim, the interpretation, the argument — originated in the machine and you adopted it, you have outsourced authorship. If it originated with you and the machine helped you express or check it, you have used a tool.
• The disclosure test. Would you be comfortable describing exactly how you used AI, to your supervisor, in writing? If describing it feels like confessing, you already know.
• The verification test. Every fact, citation, and quotation that passed through an AI must be verified against the primary source. Models fabricate confidently; a hallucinated citation in a thesis is fully your responsibility.

Do universities actually allow this?

Policies vary by institution and are moving fast, which is why the practical rule is: your university's current written policy and your supervisor's explicit guidance override any general advice — including this article. What is consistent across serious institutions is the direction: transparent, disclosed, tool-level use is being accommodated; undisclosed generation of academic work is misconduct everywhere.

What does responsible disclosure look like?

Simple and matter-of-fact — a methods-adjacent note: which tools, for which tasks, with what verification. "AI tools were used to accelerate literature discovery and to improve the clarity of author-written text; all sources and claims were verified against originals; analysis and conclusions are the author's own." Adjust to your institution's format. Hiding use is what converts acceptable practice into an offence.

Why learn this in a structured way?

Because the difference between a researcher who saves ten hours a week and one who risks their degree is not the tool — it is the workflow around it. That workflow is exactly what FuKazee's ScholarShift Accelerator teaches: a 4-week evening cohort for Masters and PhD students covering literature review, citation systems, methodology support and academic writing, with integrity rules taught explicitly, not as an afterthought. Cohorts have drawn researchers from the University of Nairobi, JKUAT, Kenyatta University and beyond.

The framework on one page

• AI for discovery, comprehension, organisation, language — yes
• AI for claims, analysis, argument — no
• Verify everything against primary sources
• Disclose use plainly; your institution's policy wins

Want the full workflow, taught live? See the ScholarShift Accelerator or start with our guide to the best AI tools for postgraduate studies.