Perplexity finds real academic papers and shows you exactly where they are. ChatGPT writes text that looks like citations but often points to papers that don't exist, and Nigerian supervisors have started checking every single one.
If your supervisor has ever asked you to produce a paper you couldn't find, this is what happened to you. Here's why it keeps happening and what to use instead.
Why ChatGPT keeps getting Nigerian students in trouble
ChatGPT is not a search engine. It's a language model trained to produce text that sounds right. When you ask it for academic references, it generates something that looks exactly like a citation: author surname, initials, year, journal name, volume number, pages. Properly formatted, confident, completely convincing.
But it is not retrieving those papers. It is predicting what a citation in this field would look like, based on patterns in its training data. The paper it describes may never have been published. The author may exist but never wrote that study. The journal is real but that article is not in it.
A study published in Scientific Reports found that more than half of ChatGPT's academic citations were either fabricated or contained significant errors. On niche or regional topics (exactly the kind of research a Nigerian student writing about African markets, local business behaviour, or Nigerian public institutions would be doing) fabrication rates climb to 30 to 55 percent. The latest GPT models have the same problem. It has not been fixed.
Your supervisor knows this now. Google Scholar searches during supervision sessions are common. Getting caught doesn't just mean resubmitting. Some students are facing academic misconduct charges for work they genuinely believed was legitimate.
What happened to Chisom at LASU
Chisom was a 400-level Mass Communication student with two weeks to strengthen Chapter 2 before her next supervision session. She typed her research question into ChatGPT: "Give me academic sources on the influence of social media on political participation among Nigerian youth."
Six references came back. Properly formatted. Nigerian-sounding author names. Journals she recognised, or thought she recognised. She cited all six and submitted.
Supervision day. Her supervisor opened Google Scholar mid-session. First citation: nothing. Second: nothing. Third: a real journal, wrong article entirely. "Chisom, I can't find any of these. Where did you get them?"
She had no answer. The session ended early. Resubmit.
Her coursemate Emeka had used Perplexity for the same topic. He ran Academic Focus first for a quick scan, then used Deep Research for his core question, which took about four minutes. He got a structured report with 43 citations, every one a clickable DOI link. His supervisor found every paper. Chapter 2 passed that day.
What Perplexity actually does
Perplexity searches the real web in real time. When you ask it a research question, it retrieves actual pages, reads them, and summarises what it found, with numbered citations linking directly to the source. Every claim has a number. Every number goes to a real page you can open.
It is not generating text that looks like a reference. It is finding the reference and showing you the address.
Two features on the free tier are especially useful for final year students:
Academic Focus
Before you type your query, click Focus and select Academic. This restricts results to peer-reviewed databases. Use this for targeted searches: finding one specific paper, checking what researchers say about a narrow concept, or following up on something your supervisor mentioned. Results are fast and each one is verifiable.
Deep Research
This is the one most students don't know about, and it's free, with 5 queries per day. When you click Deep Research before submitting, Perplexity runs for 2 to 4 minutes doing dozens of searches automatically. It comes back with a structured multi-section report covering your topic, with analysis and 40 to 50+ citations. Think of it as a research assistant who spent an hour in the library, except it took four minutes and every source is clickable.
For your literature review, Deep Research is the right tool. Use your 5 free queries on your core research questions, not on things a basic search can handle. "What does existing research say about the effect of social media on political participation among Nigerian youth?" is a Deep Research question. "Who is the author of X paper?" is not.
How to use it
Go to perplexity.ai. No account required. No dollar card. No VPN. Everything below works on the free tier.
For a targeted search: click Focus, select Academic, ask your question in plain language. Not keywords, a proper question. "What do researchers say about media influence on youth voter turnout in sub-Saharan Africa?" returns better results than "social media youth Nigeria voting."
For your literature review: click Deep Research, ask your full research question, wait 3 to 4 minutes, read the full report. Go through every citation it surfaces. Click the link. Confirm the paper exists. Skim the abstract. If it's behind a paywall, search the exact title on Google Scholar first; many authors upload free versions. The Unpaywall browser extension also unlocks a large number of paywalled papers automatically.
Cite the original paper in your work, not Perplexity. Your supervisor wants the journal article, not the tool you used to find it.
The mistake that undoes everything
Treating Perplexity like ChatGPT: asking it to write your literature review and pasting the output directly into Chapter 2.
Perplexity's summaries and Deep Research reports explain what papers argue. They are not your words. Copying them into your project is plagiarism, and the sources being real doesn't change that.
Use Perplexity to find papers and understand their arguments. Write the actual sentences yourself. ChatGPT still has a role here: structuring your argument, simplifying dense academic language, helping you draft non-research sections. Just not for finding sources.
Frequently asked questions
Can I use Perplexity for free in Nigeria?
Yes. No payment, no dollar card, no VPN. The free tier includes unlimited standard searches, 5 Deep Research queries per day, and Academic Focus. Go to perplexity.ai and start.
Will my supervisor accept papers I found through Perplexity?
Yes, because you cite the original journal paper, not Perplexity. Your supervisor sees a real paper with a real DOI. They won't know how you found it, and it doesn't matter.
Is Perplexity always accurate?
No. It occasionally returns a broken link or slightly wrong information. Always click each citation and confirm the paper exists before using it. This is especially true for very recent publications and niche Nigerian topics.
What if the paper is behind a paywall?
Search the exact title on Google Scholar first. Many authors upload free versions. The Unpaywall browser extension also unlocks a large number of paywalled papers automatically and is free to install.
Does ChatGPT still fabricate citations in 2026?
Yes. Even the latest GPT models do this. The problem is worst on niche and regional topics, exactly what most Nigerian final year students are writing about. Don't use ChatGPT to find sources.