The Mindset
Before you touch a single tool, get the mindset right. AI is a thinking partner, not a citation engine. The quality of your literature review depends more on how you use it than which model you choose.
"AI is a thinking partner, not a citation engine."
Everything in this module follows from that one sentence.
From passive to intelligent user
Don't ask AI to do the thinking. Ask it to surface, structure and challenge what you already half-know. The intelligent user steers; AI accelerates.
Question-first, not answer-first
Frame every prompt as a hypothesis or a structured question. A vague question yields a vague paragraph. A precise question yields a usable paragraph.
Source-grounded synthesis
Never trust an AI summary without the PDF in hand. The model's job is to help you read faster — not to read for you. Verify every claim, every citation.
Iterative dialogue (the Sandwich)
Context → ask → critique → refine. Treat AI like a sharp but forgetful colleague: it does its best work when you give it tight context and push back hard.
Passive consumer vs. intelligent user
The same tool produces wildly different results depending on who's at the keyboard. The intelligent user brings their own framing, sources and judgement.
The Sandwich Method
Wrap every AI interaction in human context and human critique. Your judgement is the bread; the AI is the filling.
Red flags to watch for
- Hallucinated DOIs and authors — models invent plausible-looking citations. Always verify in Scopus / Google Scholar.
- Outdated training data — most chat models lag the literature by 6–18 months. Use search-enabled modes for recent work.
- Confident summaries of papers it never read — if you didn't upload the PDF or give it a working link, the summary is a guess.
- Flattening — AI tends to smooth out disagreements between authors. Ask explicitly: "Where do these papers disagree?"