Class 9 · Unit project
Unit 4 project: Make it, check it, own it
Use generative AI to help make something real for your school. Then prove every fact in it — and prove you can explain it with the screen shut.
The task
This unit taught four habits: ask well, check everything, say honestly how it was made, and be able to explain it yourself. Today your group uses all four on something real that the school could actually use.
Only your teacher operates the AI account, on the projector. Your group never signs up for or types into an AI tool itself, and no prompt contains anybody's personal information.
Work in groups of 3–4. One period. Everything in own-it.odt.
Choose a piece (3 min)
Pick one, about 120–150 words:
- A notice for parents about a real school event, with
[DATE]and[PLACE]left blank - A one-page explainer for Class 6 on a topic from this year's science or geography
- A visitor's note about a local place — a shrine, a lake, a garden, a market — for guests at the school's exhibition
Make it (12 min)
- Prompt one — write it with all four parts: role · task · context (including do not invent…) · format. Check it for personal data. Your teacher runs it.
- Log the answer, and what's wrong with it.
- Prompt two — fix the prompt, not the answer. Your teacher runs it. Log it.
Check it (12 min)
- Split the second answer into claims, one per row: claim · verdict (supported / contradicted / unverifiable) · named source.
- Keep only what you could support. Rewrite the piece in your own words — the tool's draft is a starting point, not your text.
Own it (8 min)
- Under the piece, write the AI-use line: which tool · what it did · what you did · what you checked, against what.
- The four-minute test, alone. A student from another group asks you two questions about your piece with the screen shut. They note what you could and couldn't explain.
How it is marked
Each row is marked 0–3. Rows 1–5 are shared by the group; row 6 is yours alone. Marks are returned privately and never compared with anyone else's.
Total: 18. A shorter piece with every fact checked scores higher than a long one with none.
1. The prompts
- 0Missing
- 1One vague prompt
- 2Two prompts with all four parts
- 3Two prompts, the second fixing a named problem, no personal data
2. The prompt log
- 0Missing
- 1Answers pasted, nothing judged
- 2What was wrong, noted
- 3What was wrong, and how prompt two fixed it
3. The claim check
- 0Missing
- 1Some claims, no sources
- 2Every claim with a verdict
- 3Every claim, a verdict and a named source
4. The final piece
- 0Missing
- 1The tool's text, lightly edited
- 2Rewritten, only supported claims
- 3Rewritten in the group's voice, useful to its reader
5. The AI-use line
- 0Missing
- 1"We used AI"
- 2All four parts
- 3All four parts, precise enough to repeat
6. Screen shut
Just you- 0Couldn't explain
- 1Explained one answer
- 2Explained both
- 3Explained both, and said what they were unsure of