Library › Research AI › College students

Check a chatbot's citations against your library search

Practice: Catching mistakes and made-up facts. About 5 minutes.

About this exercise

A chatbot gave you 5 citations for your Psychology 230 paper on phones and sleep, and you searched each one in the library database. You want exactly 5 numbered lines, one per citation, each labeled Found, Wrong details, or Not found, and each Wrong details line must name the detail that differs. Base the labels only on your library results. Plain text, nothing else.

Common mistakes when you ask AI for this

  • Asking "are these real?". The AI says it can't browse and gives general tips instead of using your search results. Try: Say: "Label each one using only my library results below."
  • Counting a near match as found. The AI sees the same author and year for #2 and calls it close enough. Try: Define it: "Found means title, authors and journal all match."
  • Vague 'wrong details' notes. The AI says a citation "has errors" without saying which, so you can't fix it. Try: Add: "For Wrong details, name the field that differs."
  • Not saying how many. The AI decides the length itself and usually returns more lines, one per citation than you can use. Try: Be exact: "exactly 5 numbered lines, one per citation."
  • Getting markdown back. AI defaults to bold labels and headings, which show up as stray symbols in your reference notes. Try: Ask for "plain text, no bold, no headings."

What a strong prompt includes

  • Gives exact labels
  • Uses your search as the judge
  • Asks which detail differs
  • Says exactly how many
  • Asks for plain text

Practice this now ↑

Related exercises