01 · The tools
Which AI tools to use for a literature review
Elicit
VisitFinding relevant papers and extracting findings, methods and sample sizes into a comparable table
Consensus
VisitSeeing what the weight of real studies says on a specific question, with links to each paper
Semantic Scholar
VisitSearching the academic literature and following citation trails to related work
NotebookLM
VisitUploading the papers you have gathered and asking questions grounded only in those documents
02 · The method
Step by step
- 1
Search with research tools, not chatbots
Use Elicit, Consensus, Semantic Scholar or your library to find papers. These return real, openable sources rather than invented ones.
- 2
Build a source table
Ask Elicit or a spreadsheet to capture each paper's question, method, sample and finding. This is the backbone of a synthesis, not a summary.
- 3
Read the key papers yourself
AI summaries miss nuance and sometimes reverse a finding. Read the studies you will lean on so your review is accurate.
- 4
Ask AI to map themes, not write the review
Give it your verified notes and ask it to group studies into themes, agreements and contradictions. Use that as scaffolding for your own writing.
- 5
Verify every citation and claim
Confirm each source exists, and that your description of its findings matches the paper. Never cite something you have only seen AI describe.
03 · Use this now
Copy-paste prompt to synthesize themes
You are helping me synthesize a literature review. Here are my notes on the papers I have read, each with the finding I recorded: [paste notes with authors and years]. Group these studies into themes, and within each theme note where they agree, disagree or leave a gap. Use only the papers and findings I gave you. Do not add any study, citation or statistic I did not provide. Output a thematic outline I can write from, noting which papers support each theme.
04 · Avoid these
Common mistakes
- Citing papers you found only through an AI summary. It can invent sources or misstate findings; read and verify each one.
- Treating an AI summary as the paper. Summaries drop nuance and occasionally flip a result, which corrupts a review.
- Asking a general chatbot to 'write a literature review on X', which produces confident, unverifiable and often fabricated references.
05 · Questions
Frequently asked questions
What is the best AI for a literature review?
For finding and summarizing real papers, Elicit and Consensus are purpose-built, and Semantic Scholar is strong for citation trails. Use ChatGPT or Claude only to organize themes across sources you have already read and verified.
Can AI write a literature review for me?
It can help structure and draft, but you should not rely on it to write one unsupervised. It misrepresents findings and invents citations, and the review must accurately reflect real sources you have read.
How do I avoid fake citations in a literature review?
Find sources in dedicated research tools or your library, not a general chatbot. Open every reference, confirm it exists, and check that your summary matches what the paper actually reports.