Digital Technologies & AI Ethics • Unit 7 • Years 9–13 • Skills

Prompt Engineering 101

Clear, specific prompts can help an AI produce a more useful response, but prompt wording is only one factor. Results also depend on the model, its training data, supplied context or tools, and generation settings; factual claims still need checking. This handout teaches four useful prompting pillars — and why cultural awareness is part of the skill.

Ingoa / Name
Akomanga / Class

Best for

Unit 7 mid-unit skills lesson — after students have the foundational understanding of LLMs from the introductory handout. Use this before any AI-assisted project work.

Kaiako use

Model the technique on a projector before students practise. Run the "Prompt Makeover" activity as a whole-class warm-up first, then release to individual practice. Debrief the cultural sensitivity section explicitly — it's not optional.

Ākonga use

Read the four pillars, then attempt the prompt makeover section before looking at the examples again. The ethics checklist at the end is a tool you should use every time you prompt an AI — not just in class.

Free skills handout, premium localisation path

Want this adapted to approved classroom tools and subject tasks? Te Wānanga can change format and difficulty. Any NZ-history or te reo Māori example must remain attached to teacher-supplied, verified sources; the tool must not generate local authority or authenticate translation.

  • Tool-specific prompt examples for the AI tools your school has approved.
  • Prompting tasks for Social Studies and Science contexts.
  • Source and permission checks for any te reo Māori or Māori-topic prompt.

Kaiako planning snapshot

  • Use length: 45–55 minutes. Concept read-through (~15 min), modelled examples (~10 min), prompt makeover practice (~20 min).
  • Grouping: Pairs for the prompt makeover activity — discuss before writing. Individual for the ethics checklist application.
  • Prep: A live AI tool projected helps greatly for demonstrating the before/after difference of prompt quality. If not available, the handout examples are sufficient.
  • Differentiation: Entry: complete pillar 1 (specificity) only and rewrite one weak prompt. On-level: use all four pillars and rewrite both practice prompts. Extension: write three original prompts for a real project task and evaluate each against the ethics checklist.
  • Neurodiversity support: The four pillars work well as a checklist. Allow students to annotate prompts rather than rewriting — circling what's missing. Sentence starters available: "I need you to act as…", "I am a Year [X] student who needs…"
AI literacy Digital citizenship Ethical reasoning

Resources already provided

  • Four pillars of effective prompting with before/after examples
  • Source, scope, permission, and authority checks for Māori or Indigenous topics
  • Two prompt makeover exercises with response spaces
  • Pre-prompting ethics checklist

All skills content and practice exercises are provided. A live AI tool enhances the lesson but is not required.

Ngā Whāinga Akoranga / Learning Intentions

  • We are learning to craft useful AI prompts using specificity, context, task role, and examples.
  • We are learning that a better prompt can improve format or relevance but cannot create factual or cultural authority.
  • We are learning to attach Māori or Indigenous claims to named sources and keep restricted knowledge out of AI systems.

Paearu Angitu / Success Criteria

  • I can explain what each prompting technique changes and what it cannot guarantee.
  • I can rewrite a weak prompt into a clearer one using at least two techniques.
  • I can identify the source, scope, permission, and human-review requirements before prompting about Māori language, history, or knowledge.

He tapatahi o te pātai / The art of the ask

Prompt wording shapes an AI system’s output, but fluency is not accuracy and specificity is not authority. A careful prompt states the task, audience, format, evidence boundary, and uncertainty rule. For Māori or Indigenous topics, begin with a named, public source and state its iwi, hapū, place, or organisational scope. Do not enter restricted knowledge or ask the model to invent a cultural position.

Ngā pou e whā / The four pillars of effective prompting

Each pillar makes your prompt more specific and useful. Use as many as fit your task.

1. Be specific and clear

Weak: "Write about dogs."
Better: "Write a 100-word paragraph about golden retriever loyalty, in a warm tone suitable for Year 7 students."

2. Provide context

Weak: "Explain the Treaty."
Better: “I’m a Year 10 student comparing solar and wind generation. Explain two advantages and two limitations of each in a table, and state which claims need a current source.”

3. State the task role

Weak: “Improve my paragraph.”
Better: “Act as a copy editor. Improve this paragraph’s clarity for a Year 10 reader without adding new facts. List every factual claim that still needs a source.”

4. Use examples (few-shot prompting)

Weak: “Sort these notes.”
Better: “Follow this pattern: ‘The policy changed in 2024’ → claim; ‘Agency report, p. 6’ → evidence. Label each line in my notes as claim, evidence, or question.”

Source and authority boundaries

Prompt framing can affect an output, but it cannot repair missing data, authenticate te reo Māori, or supply cultural authority.

Māori or Indigenous topics

Supply a named, public source from the relevant authority. Ask the model to use only that source, preserve its iwi, hapū, place, or organisational scope, cite each claim, and mark anything else “Not verified”.

Te reo Māori

Do not treat AI translation as authoritative. Use language approved for the task by a qualified speaker or named source, and require qualified human review before publication.

Contested histories

Name the sources and the people or institutions whose positions they record. Ask the model to distinguish quoted evidence, interpretation, disagreement, and missing authority; never request one generic “Māori perspective”.

Whakarite kupu / Prompt makeover

Rewrite these weak prompts using the techniques you've learned. Aim to use at least two pillars in each rewrite.

Weak prompt 1:

"Tell me about climate change."

Which pillars will you use, and why?

Your improved prompt:

Weak prompt 2:

"Write a story."

Which pillars will you use, and why?

Your improved prompt:

Āta whakaaro / Ethics checklist — before you prompt

Run through this before using an AI prompt for any real task. Check each box honestly.

  • Do I have a named source and appropriate permission for any cultural or te reo Māori material, and is it permitted to enter this AI system?
  • Could my prompt lead to biased or harmful outputs? Have I tried to prevent this?
  • Am I being transparent about using AI assistance in this task?
  • Does my prompt encourage me to think critically rather than just accept what the AI produces?
  • Have I considered data sovereignty — what information am I sharing, and with whose systems?

Entry, on-level, and extension pathway

Entry

Learn pillars 1 and 2 (specificity and context). Rewrite prompt 1 only using these two pillars. Complete three items from the ethics checklist.

On-level

Use all four pillars. Rewrite both prompts, each using at least two pillars. Complete the full ethics checklist and discuss your answers with a partner.

Extension

Write three original prompts for a real upcoming assessment or project. Evaluate each against all four pillars and the ethics checklist. Identify which pillar made the biggest difference and explain why.

Aronga Mātauranga Māori

Prompt wording is not cultural authority. Te Mana Raraunga’s published principles state that Māori have authority over Māori data and that all data has whakapapa, including provenance, purpose, context, and the parties involved. A responsible prompt can require a named source, preserve its scope, protect restricted material, and expose uncertainty. It cannot turn an AI persona into a historian, translator, kaumātua, or representative of Māori.