Adapted from the Course Pack v1.0 (sources/course-ai-ethics-course-pack.md), which carries full scripts and talking points. This guide is the operational layer: adjusted run sheets (S2/S3 reordered case-first per syllabus §3.1), deck cues, and logistics. Decks: 02-slides/s1–s7.html. Prompts: prompt-handouts.md. Instruments: 04-assessments/.
Threading rule: every session ends by loading the next one. The course is one argument in seven movements, resolved in an action plan.
| Time | Block | Deck / material |
|---|---|---|
| 0:00–0:10 | Cold open: the 90-second assignment (P01, live) | s1 #2–3 |
| 0:10–0:20 | Framing: the gap is the curriculum | s1 #4 |
| 0:20–0:35 | EAI-CMM baseline (honesty guard) | s1 #5 · eai-cmm.md ×1 each |
| 0:35–1:00 | Data walk: state of AI 2026 | s1 #6–13 |
| 1:00–1:20 | Opportunity ledger + Learning Mode demo | s1 #14–18 |
| 1:20–1:50 | Threat/Gift Matrix (A2 charts) | s1 #19–20 |
| 1:50–2:00 | Exit ticket 3-2-1 + contrarian close | s1 #21–22 · D1 cards |
Notes: capstone brief handed out here (participants know the destination from hour one). Teams formed now are fixed for three days. Master chart stays on the wall.
Reordered case-first — deviates from pack run sheet by design.
| Time | Block | Deck / material |
|---|---|---|
| 0:00–0:05 | Re-entry: the natural experiment | s2 #2 |
| 0:05–0:20 | CASE: Ofqual 2020, worked cold | s2 #3–5 · dossier |
| 0:20–0:30 | Work the case (tables; harvest exact words on board) | s2 #6 |
| 0:30–0:40 | Anchor: St George's 1980s — same failure, 40 years apart | s2 #7–8 |
| 0:40–0:50 | Extraction: principles named as labels for what the room said | s2 #9–10 |
| 0:50–1:05 | Live bias demo (P03) + local framing | s2 #11–14 |
| 1:05–1:20 | Risk map: six categories | s2 #15–16 |
| 1:20–1:50 | Ethics Triage (8 cards, strict ranking) | s2 #17–18 · cards in templates.md |
| 1:50–2:00 | Responsibility triad + case memo assignment + close | s2 #19–21 · D4 |
Do NOT show slide #10 (five principles) before the room has worked the case. If the P03 run comes back clean, use the scripted pivot (s2 #13) — the audit-per-model lesson is stronger than the gotcha.
Reordered case-first — Texas A&M opens before any detector data.
| Time | Block | Deck / material |
|---|---|---|
| 0:00–0:10 | Case memos ×3 + exit-ticket answers | s3 #2 |
| 0:10–0:25 | CASE: Texas A&M–Commerce, worked cold | s3 #3–6 · dossier |
| 0:25–0:35 | Reframe: what plagiarism was for | s3 #7 |
| 0:35–0:55 | The detector evidence (deliver slowly) | s3 #8–11 |
| 0:55–1:25 | Detector on Trial (moot, steelman twist) | s3 #12–13 |
| 1:25–1:40 | Successor regime + P05 live disclosure draft | s3 #14–16 |
| 1:40–1:55 | Copyright fault lines + the misuse conversation | s3 #17–18 |
| 1:55–2:00 | Warm-up ticket + contrarian close | s3 #19–20 |
| Time | Block | Deck / material |
|---|---|---|
| 0:00–0:15 | The 4Ds in twelve minutes | s4 #2–3 |
| 0:15–0:30 | Facilitator models the loop (P07 thin→full, narrated) | s4 #4 |
| 0:30–1:05 | Lab 1: Rubric builder | s4 #5–6 · D2 log |
| 1:05–1:40 | Lab 2: Feedback assistant (anonymize first — PDPA) | s4 #7–8 |
| 1:40–1:55 | Gallery: three logs projected | s4 #9 |
| 1:55–2:00 | Bridge to Part 2 | s4 #10 |
Coaching rule: never touch keyboards; ask "what's wrong with this output?" — articulation trains Discernment.
| Time | Block | Deck / material |
|---|---|---|
| 0:00–0:20 | The crutch effect (the one study to remember) | s5 #2–5 |
| 0:20–0:30 | Seven roles + three lanes | s5 #6–7 |
| 0:30–1:10 | Lab 3: Redesign sprint (audit → lanes → swap-test) | s5 #8 · D3 |
| 1:10–1:40 | Lab 4: Guardrailed tutor + break-test | s5 #9 |
| 1:40–1:55 | Lab 5: Policy paragraph (first-year test) | s5 #10 |
| 1:55–2:00 | Portfolio check + exit word + contrarian close | s5 #11–12 |
| Time | Block | Deck / material |
|---|---|---|
| 0:00–0:10 | Re-entry: from craft to constitution | s6 #2 |
| 0:10–0:30 | Institutional maturity scan (10 items; harvest on board) | s6 #3 · eai-cmm.md §institutional |
| 0:30–0:50 | Policy anatomy (7 components) + national frame | s6 #4–5 |
| 0:50–1:05 | Crosswalk: Bill principles ↔ S2 lens; risk tiers ↔ lanes | s6 #6–7 |
| 1:05–1:15 | What good looks like (default rules; teachable-policy test) | s6 #8 |
| 1:15–1:45 | Gap analysis (P12; cost-of-inaction mandatory) | s6 #9 |
| 1:45–2:00 | Drafting brief (gate) + close | s6 #10–11 |
Bill language: always "proposed"/"consultation." Verify Bill + MOHE status the week of delivery.
| Time | Block | Deck / material |
|---|---|---|
| 0:00–0:05 | Sprint rules | s7 #2 |
| 0:05–0:50 | Drafting sprint (P13; three interventions only) | s7 #3 · D5 skeleton |
| 0:50–1:15 | Red-team exchange (P14; written findings only) | s7 #4 |
| 1:15–1:30 | Patch round (fix 3, backlog the rest) | s7 #5 |
| 1:30–1:40 | EAI-CMM retake + delta | s7 #6 · eai-cmm.md ×1 each |
| 1:40–1:55 | Action plan skeleton (P15; kill criterion mandatory) | s7 #7 · D6 |
| 1:55–2:00 | Contrarian close: the last axiom | s7 #8–9 |
7 min each (5 + 2 panel Q). Panel: facilitator + institutional leader + rotating peer judge. Rubric in 04-assessments/templates.md. Pass ≥12 · Distinction ≥17 · one "Ship It" award.
Facilitator kit: projector + spare HDMI/USB-C · two AI accounts pre-logged (primary + backup vendor) · offline screenshots of every live demo (P01, P03, Learning Mode, P05) in deck appendices · printed EAI-CMM ×2 per participant · scenario cards ×8 per table · A2 flip charts + markers · index cards · visible timer.
Per participant: laptop + working AI account (verify at daftar masuk) · one real course's materials (mandatory pre-work) · printed prompt handouts + templates.
Room: tables of 4–5, mixed-discipline, fixed 3 days · wall space for the Threat/Gift master chart · Wi-Fi stress-tested for 30+ concurrent AI sessions.
Pre-course email (T-7 days, i.e., by 25 Aug): account setup instructions · bring-a-real-course instruction · optional primer: AI Fluency Framework & Foundations (free, ~3–4 h).
Language: deliver in English or BM as the room prefers; artifacts drafted in the language the institution's policies actually use.