TFIS · AI Ethics in Education · Course
Hands-on Workshop:
Ethical AI Tools — Part 1
Teaching, Assessment and Student Learning
Session 4 · Wednesday 2 September 2026 · 11.00 am–1.00 noon
The 4Ds in twelve minutes · AI Fluency Framework
| Competency | The move |
|---|
| Delegation | The decision before the prompt: should AI touch this task, in what role? Educator heuristic: delegate production, never judgment. Grades, admissions, integrity findings — AI is a second reader at most. |
| Description | Prompting as professional communication, not incantation. Six moves: context, examples, constraints, steps, think-first, role & tone. If you can brief a research assistant, you have this skill — you've been under-briefing the model. |
| Discernment | Judging what comes out — three layers: product (correct? complete?), process (sound reasoning or fluent mush?), behavior (drifting, over-agreeing, padding?). |
| Diligence | Owning the output: verify before it touches students, disclose how it was made, protect data going in. The D that makes the other three ethical rather than merely effective. |
Dakan, Feller & Anthropic · CC BY-NC-SA — you may legally remix these materials for your own courses; this course does exactly that
This morning's engine
Describe → generate → discern → re-describe
Fluency is the loop run fast — not the first prompt written well.
The remaining pair — Delegation↔Diligence — is this afternoon's spine.
Live build · facilitator models · 15 min · P07
Watch the loop, narrated
- Deliberately thin prompt: "make me a rubric for an essay." Show the generic mush.
- Rebuild with full Description: course, level, learning outcome, band descriptors, local grading scale. Show the difference.
- Discern aloud: "Criterion three overlaps criterion one — that's padding. The 'credit' band isn't observable behaviour — rewrite." Two more iterations.
SAY
"What you just watched is the entire skill. Not the prompt — the loop. Now you run it."
Lab 1 · Rubric builder · 35 min · individual, coached
A real assessment. This semester. No hypotheticals.
- Run P07 with full context → iterate minimum 3 rounds.
- Each round: log one row in the Verification Log — what I asked · what was wrong/weak · what I changed.
| Quality bar (stays on screen) |
|---|
| Every criterion observable | Band descriptors distinguishable by a colleague |
| Aligned to a stated learning outcome | Local grade-scale compliant |
Coaching pattern · facilitator
Never touch keyboards. Ask: "What's wrong with this output?"
Make them name it. Discernment is trained by articulation, not correction.
Lab 2 · Feedback assistant · 35 min · higher stakes
Non-negotiable setup rules
- Use a past, anonymized student excerpt — names, IDs, identifying details stripped before anything is pasted.
- This is EAI-CMM item 17 being practiced. In Malaysia it has a statute behind it: PDPA 2010.
- Output touches students directly — Diligence rules bind.
Lab 2 · the spec · P08
AI drafts. You author.
- Model = feedback drafter against your Lab-1 rubric: (a) three specific strengths quoting the text, (b) three growth points phrased as questions, (c) one concrete next step. Explicitly not a grade — Delegation boundary.
- Discern for: hallucinated praise of things the text doesn't do · feedback the student can't act on · tone mismatch with your voice.
- Finish by editing the draft into your voice.
SAY
"The feedback that reaches the student is yours. The model drafted; you authored.
That distinction is the whole ethics of this lab."
Gallery · 15 min · three screens
Project the logs — not the outputs
- Three volunteers project their verification logs. The room inspects the iteration path.
- Formative check, collected: log with ≥3 rows + one sentence: "The most useful thing I changed between round 1 and round 3 was ___."
- A log with real deltas = this session's objective, met.
Bridge to Part 2
SAY
"You now have leverage — two workflows that give you hours back. This afternoon we spend those
hours where they matter most: on the assessments themselves, and on the hardest question in this
whole field — proof that AI can raise your students' scores while lowering their learning.
Makan dulu; come back dangerous."
Appendix · facilitator only
Fallbacks & logistics
- [ insert screenshots: P07 thin-vs-full comparison, 3 iterations ]
- Wi-Fi stress point: 30+ concurrent AI sessions. Fallback: pair sharing on alternating machines.
- Handouts: prompt library P07–P08 + verification-log template (D2), printed per participant.