Research with AI

Where AI helps in research, and where it doesn't

AI is a strong starting point for research, and a weak finishing point.

The rule: AI points you to the research; it doesn't replace it.

Start with an overview and search terms

I'm writing a 2,000-word essay on the effects of mobile money on small
businesses in Nigeria. Give me: a short overview of the topic, the main
debates, and 10 search terms I could use on Google Scholar.

Now you know what to look for. Use those search terms in Google Scholar, your school library portal, or trusted websites, not the AI chat.

Never trust an AI reference

AI assistants can invent references: realistic-looking author names, journals and page numbers for articles that don't exist. Submitting one can get you penalised for fabrication.

For every source you use:

1. Find it yourself: search the exact title on Google Scholar or your library.
2. Open it and check that it actually says what you're claiming.
3. Cite the real source, not the AI.

Some assistants can search the web and show links (for example ChatGPT's search, Gemini, Copilot, or Perplexity). That's better, but still open each link: summaries can misstate what a page says.

Summarise and question real sources

Once you have a real article (a PDF you're allowed to use, or a web page), AI becomes very useful:

Here is an article I'm using for my essay. Summarise its main argument,
the evidence it uses, and any limitations the authors mention.

Compare these two articles. Where do they agree and disagree?

Then read the key parts yourself. The summary saves time; it doesn't replace understanding.

Try it

1. Choose a topic for an assignment you have now, or one you'd like to explore.
2. Ask an AI assistant for an overview and 10 search terms.
3. Use three of those terms in Google Scholar and find two real articles.
4. Summarise one article with AI, then read its conclusion yourself and compare.
5. Ask the AI for "three references on this topic", then try to find each one. How many are real?