Responsible use

Ethics, Accuracy & Academic Integrity

Everything in the previous modules produces material that goes out under your name to learners. This part is the professional discipline that makes that safe: check it, protect student data, say when AI helped, and design assessment that still means something.

Session: Day 1 · 3.15 – 4.45 pm
Focus: Accuracy · privacy · integrity
Outcome: Your own classroom AI policy

Fact-checking and quality control

Generated text is fluent by design and accurate only by luck. Build a fixed habit rather than relying on how confident the output sounds.

1
Ask it to flag its own uncertainty
Add “Mark anything I should verify with [CHECK] and list what you are unsure about” to every prompt.
2
Verify every specific
Names, dates, formulas, statistics, syllabus codes, citations. Invented references are the most common failure.
3
Ground the task where possible
Upload your own source and require answers from it only — NotebookLM is built for exactly this.
4
Read as a subject expert
You are the domain check. If a sentence feels slightly off for your field, it usually is.
5
Restore your voice
Rewrite at least the opening and closing in your own words so it reads like you, not like a machine.

Never use AI for a claim you cannot verify

If you cannot check it, remove it. A wrong fact taught confidently is harder to undo than a gap left unfilled.

Data privacy essentials

	Safe to put in	Never put in
Student work	An anonymised extract with names and IDs removed	Scanned scripts, name lists, class registers
Results	Aggregated, anonymised figures (Student A, B, C)	Named mark sheets or report cards
Documents	Your own notes, public syllabus documents	Confidential internal reports, exam papers under embargo
People	Generic descriptions of a learning difficulty	Named welfare, discipline or medical information

On free consumer tiers your input may be used to improve models — turn off training in settings where the tool allows it.
Prefer institutional accounts when your school provides them.
Strip identifiers before uploading, then reinsert names in your own document afterwards.
Never upload student photographs, voices or videos into avatar or voice-cloning tools.
Follow your institution's and your country's data protection rules — they override any tool's default setting.

Bias, representation and fairness

Watch for

Cultural default

Examples, names and contexts often default to Western settings. Ask explicitly for local names, currency, places and examples your students recognise.

Watch for

Stereotyped roles

Check generated images and scenarios for who appears as the scientist, the leader, the helper. Regenerate when the pattern is narrow.

Watch for

Language level drift

Output tends to be more academic than your learners need. Specify the reading level and check it against a real student.

Watch for

Accessibility gaps

Add captions, alt text and readable contrast yourself — no tool does it well by default.

Academic integrity: redesign, do not police

AI detectors are unreliable and can wrongly accuse honest students, including second-language writers. The durable answer is task design and disclosure, not surveillance.

Redesign

Assess the process

Ask for drafts, plans, annotated sources and a short reflection on what changed and why.

Redesign

Anchor in the local and personal

Tasks tied to your classroom data, a local site visit or a lesson discussion are hard to outsource.

Redesign

Add live components

A two-minute oral defence of a submitted assignment reveals understanding instantly.

Redesign

Teach AI use openly

Let students use AI, then require them to declare where and to critique its output. That is a genuine skill.

Class AI usage statement

Draft a short AI usage statement for my [level] [subject] class, under 150 words.

Cover: where AI is allowed, where it is not, how students must declare their use,
and what happens if they present AI output as their own thinking.
Tone: clear, respectful, not threatening. Written for students, not lawyers.
Give me an English version and a Bahasa Malaysia version.

Disclosure: a simple standard

On materials: “Drafted with AI assistance and reviewed by [your name].”
On AI-presenter videos: state that the presenter is an AI avatar.
On generated images or video of events: label them as illustrations, not records.
To colleagues: share the prompt as well as the output when they ask how you made it.
To students: model the disclosure you expect from them.

Hands-on task

Write your own classroom AI ground rules

0/6 steps

Goal: One page you could hand to students and to your head of department tomorrow.

Finished this part?

Mark it complete to update your progress meter and move on.