Library › Research & AI Search › Managers and team leads
Compare two AI answers about a pilot against the real report
- For Managers and team leads
- You practice Keeping the AI to your sources
- Time about 5 minutes
About this exercise
Two AI tools summarized your company's flexible-Friday pilot report differently, and you need to know which one to forward to your VP. Write exactly 4 numbered lines labeled A error:, B error:, Both missed: and Verdict:, judged only against the report. The Verdict line names A or B and must come last. Plain text, under 100 words.
Common mistakes when you ask AI for this
- Not naming the judge. The AI compares the two answers to each other or to what sounds right, instead of to the report. Try: Add: "Judge both only against the pilot report."
- Skipping what both missed. The AI only lists differences, so the shared gap (no control group) never comes up. Try: Add a line: "Both missed: what neither answer mentions."
- No clear verdict. The AI says "both have strengths" and leaves you to decide. Try: Add: "Last line: Verdict: A or B."
- No fixed shape. The AI writes a long side-by-side table you can't forward. Try: Say: "Exactly 4 numbered lines labeled A error:, B error:, Both missed:, Verdict:."
- Getting markdown back. AI defaults to bold labels and headings. Try: Ask for "plain text, no bold."
What a strong prompt includes
- Uses the report as the judge
- Names the 4 labels
- Asks for a verdict last
- Sets the length
- Asks for plain text
Related exercises
- Accept a job offer and confirm the exact terms (Writing AI · Job seekers)
- Alert sales to a competitor's price change, facts only (Research & AI Search · Product managers)
- Allocate shared rent by headcount using the right roster (Finance AI · Accountants)
- Announce a seasonal offer with its real terms and no fake scarcity (Sales AI · Small business owners)
- Answer a 'where is my order' email with the real new date (Writing AI · Customer support teams)