Participant handout
One page, printed or on your device. Keep it beside you through the session.
1. The five AI-use levels (AIAS 2.1 — icon + text, never icon alone)
| Level | Name | One-line meaning |
|---|---|---|
| 1 | No AI | Nothing generative, at any stage of this part |
| 2 | AI Planning | AI may help plan and brainstorm, before the real work |
| 3 | AI Collaboration | AI may critique and suggest during the work; the student decides, verifies, and logs |
| 4 | Full AI | AI may generate freely; the log and the student’s evaluation of it are graded |
| 5 | AI Exploration | Co-creating with AI is the task; reflection is the evidence |
Source: Perkins, Furze, Roe & MacVaugh — AI Assessment Scale, version 2.1 (levels verified on the scale’s official site, 2026-10-25; see the kit’s literature notes).
2. The two lanes (after University of Sydney)
- Secure lane (small, weighted): in-class checks, defences, vivas. Certifies the individual achievement. No AI here.
- Open lane (most of the work): practice, drafts, notebooks, feedback. AI allowed at a named level with disclosure. This is where learning happens.
Weights that work (Case B): open practice 10% · secure checks 20%.
3. The criterion-risk test
For each rubric criterion, ask: could a competent AI produce this at standard, without the learning?
- Pass — AI does it easily → redesign or reweight
- Fragile — AI approximates it → raise the bar (verification, personal context)
- Holds — needs the student’s process or live defence → keep, make visible
4. Score the sample essay (live demo)
Use your course’s normal generosity. The rubric is on the back of this sheet / the next page.
| Criterion | Round 1 (naive AI) | Round 2 (determined AI) |
|---|---|---|
| Thesis & argument (5) | ||
| Use of evidence (5) | ||
| Analysis & depth (5) | ||
| Writing & organisation (5) | ||
| Total (20) |
5. Redesign canvas
Use the canvas in the sprint. The one-line version:
What must the grade certify? → which criteria does AI pass today? → AI level per stage? → which moves? → what does the new rubric reward instead? → what does this NOT solve?
6. Where everything lives
- Kit home: Start here
- Three cases: A essay · B problem set · C research report
- Eight approaches compared — workload and equity numbers
- Decision tool — six questions, nothing stored
- Templates — syllabus + assignment AI statements, blank canvas
7. Pre/post survey
About five minutes, anonymous, five confidence items + two open questions: survey. The pre-survey comes with the calendar invite — do it before the session; the same items before and after are how we know whether the session worked.