Library › Research AI › High school students

Label each chatbot claim against your earth science text

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

About this exercise

You asked a chatbot about the 1980 Mount St. Helens eruption for Earth Science, and it gave you 8 numbered claims. You want one numbered line per claim, under 90 words total, each labeled Confirmed, Wrong, or Not in textbook, and every Wrong line must give the textbook's version. Judge only by your textbook excerpt, not by what the AI already knows. Plain text, nothing else.

Common mistakes when you ask AI for this

  • Letting the AI use its own memory. The AI checks claims against what it remembers, so it may call the Carter visit true even though your textbook never mentions it. Try: Say: "Judge only by the textbook excerpt."
  • No fixed labels. Without labels the AI writes a paragraph per claim, and you can't tell at a glance which ones to fix. Try: Give them: "Label each Confirmed, Wrong, or Not in textbook."
  • Not asking for the right version. The AI may just say "incorrect" without the textbook's number, so you still don't know what to study. Try: Add: "For each Wrong line, give the textbook's version."
  • Trusting the AI to count words. AI is bad at counting. Ask for a total word limit and it often lands 10-20% over. Try: Cap each line: "each line under 10 words."
  • Getting markdown back. AI defaults to bold labels and headings, which show up as stray symbols in your study notes. Try: Ask for "plain text, no bold, no headings."

What a strong prompt includes

  • Gives exact labels
  • Names the source of truth
  • Asks for the correct version
  • Sets a length limit
  • Asks for plain text

Practice this now ↑

Related exercises