Skills rubric for practical exercises
Use during Sessions 1 to 3 case practice and for primary pre/post performance scenarios in the pre/post check. Award one point per criterion (maximum 4 per exercise or scenario).
Criteria (all sessions)
| # | Criterion | 1 point when… | 0 points when… |
|---|---|---|---|
| 1 | Input/use decision | Appropriate green, amber, or red with a valid reason | Wrong classification or no reason |
| 2 | Source verification | At least one consequential claim checked against an authorized source | No check or wrong source |
| 3 | Bias or omission | Identifies exclusion, harmful omission, or stereotype | Misses a taught error in the case |
| 4 | Human control | Names approver, override path, or stop rule | No human accountability point |
Proficiency: 4/4 on an exercise. Series target: at least 8 in 10 participants who complete both entry and exit checks score 4/4 on the exit exercise (organizational reporting choice).
Exercise variants
Each session includes:
- One low-risk task where green (with checks) is correct
- One must-stop task where red is correct
Rejecting all AI use is not proficiency. Participants must distinguish appropriate use from prohibited use.
Facilitator scoring notes
- Score from the participant worksheet, not from eloquence in plenary
- Partial credit is not awarded; use 0 or 1 per row
- If a group disagrees, record the majority decision and note dissent in debrief
- Do not rank individuals or organizations in aggregate reports
Sample scored exercise (Session 1 SMS case)
| Criterion | Example evidence | Score |
|---|---|---|
| Input/use decision | Red: draft contradicts source and exceeds reviewer authority | 1 |
| Source verification | Compared each line to bulletin W1 at 10:00 | 1 |
| Bias/omission | Noted accessibility error (“all entrances” vs one step-free) | 1 |
| Human control | Named local communications lead for approval | 1 |
Total: 4/4
Reporting template
| Metric | Value |
|---|---|
| Participants with paired entry and exit | |
| Exit 4/4 count | |
| Exit 4/4 percentage | |
| Unpaired entry only | |
| Unpaired exit only |
Add a narrative limitations paragraph: sample size, self-selection, no control group, synthetic cases.