Library › Research AI › Parents and households

Check a viral parent-group post against the district notice

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

About this exercise

A post about a new Wednesday schedule is spreading in your town's parents' group, and you want to check it before you change your work hours. Have the AI check each of the post's 4 claims against the district's official notice only. You want exactly 4 numbered lines, one per claim in the post's order, each starting with one label (Confirmed, Wrong, or Not in notice) and then a short reason from the notice. Under 80 words, plain text, nothing else. Write a prompt that gets you exactly that.

Common mistakes when you ask AI for this

  • Asking "is this true?". The AI writes a long verdict with headings and mixes all four claims together. Try: Ask for a list: "Exactly 4 numbered lines, one per claim, in the post's order. Plain text."
  • No fixed labels. The AI says things like "partly accurate" or "misleading", which doesn't tell you what to do. Try: Say: "Start each line with one label: Confirmed, Wrong, or Not in notice."
  • Calling silence a no. When the notice doesn't mention free care, the AI often calls the claim false instead of unconfirmed. Try: Add: "If the notice doesn't mention it, label it Not in notice."
  • Not naming the referee. The AI weighs the neighbor's comment and its own guesses about schools alongside the notice. Try: Add: "Use only the Nov 4 notice as the standard. Ignore the comments."
  • Trusting the AI to count words. AI is bad at counting. Ask for "under 80 words" and it often lands well over. Try: Cap each line instead: "each line under 15 words."

What a strong prompt includes

  • Uses the notice as the standard
  • Gives exact labels
  • One line per claim
  • Limits the length
  • Asks for plain text

Practice this now ↑

Related exercises