The right role for AI
What Perplexity should and shouldn’t do in your review
Good uses
- Narrowing a broad topic into a clear question
- Finding keywords, synonyms and key authors
- Getting a quick overview of a new field
- Summarising papers you have access to
- Comparing studies and spotting possible gaps
Don’t rely on it for
- A complete or systematic search of the literature
- Paywalled papers it can’t read in full
- References you haven’t checked yourself
- Writing the review for you to submit
- Final judgements about study quality
Workflow
How to do a literature review with Perplexity in 7 steps
Define a focused research question
A sharp question keeps the search manageable. Use a framework such as PICO (population, intervention, comparison, outcome) or simply name the topic, population, setting and years.
- Ask Perplexity: “Help me narrow this topic into a researchable question.”
- Write down inclusion and exclusion criteria now.
Try: “Help me turn “social media and teenagers” into a focused literature review question with inclusion criteria.” Ask ↗Map the field with Academic search
Set the source to Academic (or ask for peer-reviewed sources only) and ask for key authors, landmark studies, main debates and common methods. This gives you a map before you dive in.
- Ask for review articles and meta-analyses first.
- Note keywords and synonyms the answers use.
Try: “Using peer-reviewed sources, what are the landmark studies and main debates on social media use and adolescent mental health since 2015?” Ask ↗Run Deep Research for a first overview
Deep Research reads many sources and writes a structured overview with citations. Treat it as a reading list and a draft map, not as your review.
- Ask it to group studies by theme or method.
- Ask it to list each source with author, year and journal.
Try: “Do a Deep Research overview of peer-reviewed studies on [topic] from 2018 to 2026, grouped by theme, with a full reference list.” Ask ↗Verify every reference
Look up each paper in Google Scholar, PubMed or your library database. Confirm the title, authors, year and journal exist and match, and read the abstract yourself. Drop anything you can’t find.
- AI tools can mis-cite or merge papers.
- Use DOIs to confirm a paper is real.
Read and extract into a synthesis matrix
Read the papers that pass your criteria. Upload PDFs to Perplexity to help extract the aim, method, sample, findings and limitations into a matrix, then check each cell against the paper.
- One row per paper, one column per question.
- Our PDF guide covers uploads in detail.
Try: “From this attached paper, extract: aim, design, sample size and setting, key findings, and limitations. Quote the sentences you used.” Ask ↗Find themes, agreements and gaps
Ask Perplexity to compare your matrix rows: where studies agree, where they conflict, and what hasn’t been studied. Then decide for yourself which gaps matter.
- Look for gaps in populations, settings, methods or time.
- Contradictions are often your most interesting section.
Try: “Based on these study summaries, where do the findings agree, where do they conflict, and what research gaps remain?” Ask ↗Write it yourself and cite the originals
Write your review in your own words, organised by theme rather than paper by paper. Cite the original studies you read, never Perplexity, and follow your institution’s rules on AI use.
- Use a reference manager like Zotero or Mendeley.
- Disclose AI assistance if your institution requires it.
Verify every reference before you use it
AI tools sometimes get author names, years or titles wrong, or blend two papers into one. A single fake or wrong reference can cost you marks or credibility. For each paper:
- Search the exact title in Google Scholar or your library database.
- Check authors, year, journal and DOI match what Perplexity gave you.
- Read the abstract (and ideally the paper) yourself to confirm it says what was claimed.
- Remove anything you can’t find. Don’t cite a paper you haven’t seen.
Template
Synthesis matrix template
Copy this layout into a spreadsheet. Fill one row per paper, then read down each column to find themes. The first row shows an example.
| Study | Aim | Design and sample | Key findings | Limitations | Theme |
|---|---|---|---|---|---|
| Author, Year | What the study set out to test | e.g. survey, 1,200 adults, one country | Main results in one or two lines | e.g. self-reported data | e.g. Theme A |
| Paper 2 | |||||
| Paper 3 |
Tip: upload each paper and use the “Extraction” prompt below to draft a row, then check it against the paper.
Free literature search builder
Enter your main concepts. You’ll get a Perplexity prompt and a Boolean search string for databases like Google Scholar or PubMed.
Prompt for Perplexity
Boolean string for databases
Prompts
Prompts for each stage of your review
Replace the words in [brackets] with your own topic.
Suggest 5 focused literature review questions on [topic], each with suggested inclusion and exclusion criteria.
List keywords, synonyms and MeSH-style terms for searching databases on [topic], and combine them into a Boolean search string.
Using peer-reviewed sources only, list the 10 most cited or influential studies on [topic] with authors, year, journal and DOI.
Find recent systematic reviews and meta-analyses on [topic] published since 2020, with their main conclusions.
From the attached paper, extract aim, design, sample, setting, key findings and limitations into a table. Quote supporting sentences.
Compare these studies’ methods and findings in a table and explain why their results might differ.
Based on the studies discussed, what populations, settings, methods or questions have not been studied enough?
Critically appraise this study’s methodology: strengths, weaknesses, bias risks and how much weight it should carry.
Tips
Tips from the research workflow
Systematic reviews and meta-analyses give you a ready-made reading list and summary of a field.
Keep your question, criteria, uploaded papers and instructions together for consistent answers.
Check the references of your best papers and the newer papers that cite them.
Note the date, prompt and databases you used. Supervisors and examiners may ask.
Get full texts through your university, then upload them for extraction.
Rules differ by university and journal. Follow them and disclose AI use where required.
FAQ
Questions about literature reviews with Perplexity
Can Perplexity write my literature review?
It can draft text, but you shouldn’t submit AI-written work as your own. Most universities treat that as misconduct. Use Perplexity to find, organise and understand sources, then write the review yourself.
Is Perplexity good for academic research?
It’s a fast way to map a field, find key papers and summarise them, especially with Academic search and Deep Research. It doesn’t replace databases like Scopus, Web of Science or PubMed for a systematic search, and it can’t read most paywalled papers.
Does Perplexity make up references?
It cites real web pages, which makes invented references less common than with chatbots that answer from memory. It can still mis-cite authors, years or titles, so verify every reference in Google Scholar or your library database.
Can I cite Perplexity in my thesis?
Cite the original papers you read, not Perplexity. If your institution requires you to disclose AI use, describe how you used it in your methods or acknowledgements.
Can Perplexity do a systematic review?
Not on its own. Systematic reviews need a documented, reproducible database search, screening by humans and a PRISMA flow diagram. Perplexity can help with scoping, keywords and summaries along the way.
Can I upload research papers to Perplexity?
Yes. Upload PDFs you have access to and ask it to extract methods, findings and limitations. Check extracted details against the paper.
Keep learning
Start your review
Map your topic in minutes
Build your search string above, run it on Perplexity with Academic sources, and verify what you find in Google Scholar.