Investigate, audit, then design
This edition uses an inquiry cycle and project checkpoints. Ākonga gather evidence about AI systems, audit who benefits or is harmed, and turn that analysis into a culturally responsive design proposal.
- Approach: Project inquiry and design studio
- Duration: Five 80–90 minute lessons plus project work
- Scaffolding: Evidence logs, audit tools, and design checkpoints
- Outcome: AI ethics analysis and a culturally responsive design proposal
A five-lesson Inquiry & Design Studio for Years 11–13, followed by project work, providing a critical and practical introduction to AI through Māori Data Sovereignty, ethical reasoning, and community-centred innovation.
Students don't just learn about AI — they interrogate it. They examine whose values are encoded in algorithms, how Indigenous data is at risk, and what genuinely Māori-centred digital futures might look like. The unit culminates in students designing culturally-responsive technology proposals for real communities.
"Mā te huruhuru ka rere ai te manu"
It is the feathers that allow the bird to fly. Knowledge and ethics are the feathers we need in the digital age.
AI systems trained on biased data reproduce and amplify that bias at scale. For Māori communities, this means technologies that misrecognise te reo, fail to reflect tikanga, and make decisions — in health, criminal justice, welfare — that disadvantage tangata whenua. Students who understand this can advocate for change. And design better alternatives.
Ngā Whāinga Ako — Learning Intentions
- Identify how AI systems encode assumptions and values, and evaluate the real-world impact of algorithmic bias on communities — particularly Māori and Pacific communities
- Explain what a named Māori data-sovereignty source states, including its scope, authority, consent, accountability, and benefit requirements
- Evaluate an existing digital technology against named evidence and identify which design or governance decisions remain with the affected authority
- Design a technology proposal that records its evidence boundary, affected groups, data permissions, governance requirements, uncertainty, and next consultation
Ngā Paearu Angitū — Success Criteria
- I can identify ethical issues within AI systems and explain their real-world impact on specific communities
- I can distinguish published evidence, AI output, my interpretation, and decisions that require cultural or community authority
- I can use a named Māori-led source accurately and keep its iwi, hapū, place, project, or organisational scope visible
- I can design a proposal with explicit consent, access, accountability, benefit, and stop/change requirements without claiming to speak for a community
Entry / On-level / Extension
- Entry: Provide worked examples of AI bias scenarios with entry-level sentence starters. Focus on Lessons 1–3. Use structured comparison tables (Western tech values vs. Māori values) before asking students to write independently.
- On-level: Complete the full 5-lesson sequence. Produce an AI ethics analysis (Lessons 2–3) and a cultural design proposal (Lessons 4–5). Include feedback cycles with peers before final submission.
- Extension: Research a specific case of algorithmic injustice affecting Indigenous communities (e.g. biased facial recognition, health-risk algorithms, welfare prediction tools). Write a submission-style report addressed to a government agency or tech company, using Te Tiriti obligations as the framework for critique.
Inclusion Guidance
- ELL / ESOL: Pre-teach key digital technology vocabulary (algorithm, bias, data, sovereignty). Allow students to discuss concepts in home language before writing in English. Many students from Pacific and Asian backgrounds will have rich perspectives on cultural technology tensions — create space for this.
- Neurodiversity: Use accessible formats with clear headings and visual supports. Neurodiverse learners benefit from structured ethical frameworks (e.g. decision trees, weighted criteria matrices) to navigate complex AI ethics scenarios. Break the design project into weekly milestones.
- Mātauranga Māori lens: Connect AI ethics to tikanga Māori values — particularly kaitiakitanga of data (who owns and controls information about Māori communities) and the principle of manaakitanga in how technologies should serve people equitably. Te Hiku Media's voice recognition work is an outstanding NZ example.
Key Competencies
- Thinking: Critical analysis of AI systems requires students to move from "does it work?" to "who does it work for, and who does it harm?" — a fundamentally different kind of thinking
- Using Language, Symbols & Texts: AI documentation, data sovereignty frameworks (CARE Principles), technical design briefs, and government submissions are all texts this unit teaches students to read and produce
- Relating to Others: Culturally-responsive design requires genuinely understanding community needs — not projecting assumptions onto them
- Participating & Contributing: The design proposal asks students to address a real community need and propose a real governance model for the technology they design
- Lesson 1: AI Foundations — How LLMs Work & What They Assume Lesson 1
- Lesson 2: Ethics & Bias — Whose Values Are Encoded? Lesson 2
- Lesson 3: Māori Data Sovereignty Lesson 3
- Lesson 4: Cultural Design — Embedding Tikanga Lesson 4
- Lesson 5: Digital Futures — Māori Sovereignty in 2050 Lesson 5
| Lessons | Lesson | Big Question | Key Activity |
|---|---|---|---|
| 1 | AI Foundations | "He aha te atamai mīharo?" | Introduction to Large Language Models and how AI systems work — and what assumptions they make. Students audit familiar AI tools (Google Search, Spotify recommendations, ChatGPT) for embedded values and cultural blindspots. |
| 2 | Ethics & Bias | "Mā wai ērā uara?" | Critical analysis of AI bias — facial recognition failures, biased hiring algorithms, health prediction disparities. Students evaluate specific cases affecting Māori and Pacific communities and use the "who benefits, who decides" framework. |
| 3 | Data Sovereignty | "Ko wai ngā rangatira o ērā raraunga?" | Te Mana Raraunga’s published principles and Papa Reo’s primary-source account. Students map provenance, authority, consent, benefit, access, and accountability without treating either source as a universal Māori position. |
| 4 | Cultural Design | "He aha te manaakitanga matihiko?" | Source-bound, authority-aware design. Students use a need identified in a named Māori-led source, separate features they can propose from decisions requiring community authority, and begin an annotated prototype with governance requirements. |
| 5 | Digital Futures | "Ko tōna āhua ake, he aha?" | Source-grounded digital-futures scenarios. Students present design proposals that distinguish evidence from speculation and state who must shape, approve, change, or stop any real implementation. |
AI Ethics Analysis
A critical analysis of one specific AI system or algorithm that affects a named community. Students identify the evidence for the bias mechanism, trace documented impacts, distinguish inference from source-backed claims, and identify any missing authority. 600–900 words or equivalent multimedia, using at least one academic or official source and any relevant source from the affected authority.
Cultural Technology Design Proposal
A design proposal responding to a need identified in a named Māori-led source. It must include the source and its scope, a problem statement, affected people, prototype or wireframe, data and consent requirements, uncertainty, and a governance map showing who may approve, change, or stop the system. Students do not invent a community mandate; they identify the consultation still required.
Te Hiku Media is the anchor example for this unit: Papa Reo’s own account describes Te Hiku Media’s Māori-led speech-recognition and natural-language-processing work beginning with te reo Māori. It says language communities should retain sovereignty over their data and receive the benefits produced from it. Under Te Hiku Media’s Kaitiakitanga Licence, data is cared for rather than owned. Keep every classroom claim attached to that source and its organisational scope.
The design proposal (Lessons 4–5) must preserve source, scope, and authority: Do not let students convert an English gloss of tikanga into a universal design rule. Require a named source for each cultural or governance requirement, distinguish student design choices from decisions reserved to the affected authority, and ask who can approve, change, stop, or withdraw data from the proposal.
AI in the classroom: If students use AI tools, require them to identify every generated claim, verify it against a named source, and disclose the use. Do not ask an AI system—or Māori ākonga—to authenticate a generic “te ao Māori context”. Restricted, personal, or locally held knowledge stays out of the system.
Facial recognition bias resources: Gender Shades (Buolamwini & Gebru, 2018) found error rates of up to 34.7% for darker-skinned women across three commercial gender-classification systems, while the maximum for lighter-skinned men was 0.8%. Read the paper’s scope with students: this was a study of gender classification, not every form of facial recognition.
📊 Aromatawai — Assessment Framework
Assessment tracks progress against this unit's Learning Intentions and Success Criteria (above) and culminates in two authentic capstones — evidence is gathered during the inquiry, not only at the end.
Formative — during the unit (Lessons 1–3). Each lesson carries its own Success Criteria; use them as checkpoints as ākonga work:
- Lesson 1 — framing the inquiry: can ākonga distinguish AI output from source-backed evidence, record scope and uncertainty, and name the authority required for any local or cultural claim?
- Lesson 2 — the bias case studies (facial recognition error rates, health-risk algorithms): can ākonga identify how bias enters AI systems and explain its disproportionate impact on Māori and marginalised communities?
- Lesson 3 — the design brief: can ākonga convert audit evidence into testable requirements while keeping community and cultural decisions with the relevant authority?
Summative — the capstones (Lessons 2–5). Mastery is shown by what ākonga build, not a separate test:
- The AI Ethics Analysis (600–900 words or equivalent multimedia) — a critical analysis of one specific AI system affecting Māori or Pacific communities, tracing the bias mechanism and documented impact, distinguishing evidence from inference, and identifying missing authority. It must use at least one academic or official source and any relevant source from the affected authority.
- The Cultural Technology Design Proposal (Lessons 4–5) — a design proposal responding to a need identified in a named Māori-led source. It includes source scope, problem statement, affected people, prototype or wireframe, data and consent requirements, uncertainty, and a governance map showing who may approve, change, or stop the system. Presented as a design critique to the class.
🚀 Whakawhānui — Extension Pathways
Three pathways of genuine depth — each extends real unit substance rather than adding busy work, and each ends in something a student produces.
- Entry — go deeper on one bias case. Choose one AI system from the Lesson 2 case studies (facial recognition, health-risk algorithms, or welfare prediction tools) and research its specific impact on an Aotearoa NZ community. Write a one-page analysis connecting the technical bias mechanism to the real community harm.
- Developing — connect data sovereignty to a real organisation. Use Papa Reo’s own account to identify what Te Hiku Media states about language technology, data sovereignty, kaitiakitanga, and community benefit. Compare those stated commitments with one other named Indigenous data-governance source, preserving each source’s scope. Present the comparison as a briefing document.
- Mastery — take it to a real audience. Write a formal submission to a government agency or technology company, using Te Tiriti o Waitangi obligations as the framework for critiquing a specific algorithmic injustice affecting Indigenous communities. Document the response and what would change in the next draft.
🔗 Te Ara Whanaketanga — Unit Progression & Next Steps
The unit moves ākonga from consumer to critic to designer. It opens by making AI visible as a system built on human choices — not neutral technology but encoded values and assumptions, examined through a source-and-authority audit (Lesson 1). From understanding, the arc turns to justice: how do AI systems encode bias, who is harmed, and what does algorithmic injustice look like when it lands on Māori and Pacific communities — through facial recognition error rates, health-risk algorithms, and welfare prediction tools (Lesson 2)? Lesson 3 asks ākonga to move from critique to requirements: converting evidence and affected-party analysis into a reviewed design brief. The final two lessons pivot from analysis to creation. Lesson 4 challenges ākonga to prototype against source-backed requirements and state which governance decisions remain with the affected authority. Papa Reo provides a named Māori-led case whose own published account keeps language technology, data sovereignty, kaitiakitanga, and community benefit connected. Everything converges in Lesson 5's source-grounded digital-futures vision — ākonga present their design proposals and reflect on what it would take to build a genuinely decolonised digital future.
Where it leads. The Cultural Technology Design Proposal is built for critique and further consultation, not automatic enactment. It gives ākonga a concrete evidence-and-governance portfolio they can carry into senior Digital Technologies work, a following unit on community action, or a revised proposal for an appropriate audience. The unit's real endpoint is a shift in how rangatahi see the digital systems around them — not as given, but as designed, and therefore as designable on different terms.
📋 Kaiako Snapshot — Success Criteria & Differentiation
Ngā Paearu Angitū — Success Criteria
- ✅ I can identify ethical issues within AI systems and explain their real-world impact on specific communities
- ✅ I can apply a te ao Māori lens to evaluate digital technologies and their effects on tangata whenua
- ✅ I can articulate what digital sovereignty means and why it matters — using examples from Te Hiku Media, Māori health data, and iwi governance
- ✅ I can design a technology proposal that embeds Māori values into its architecture, not just its interface
Differentiation & Inclusion
Entry-level learners: Provide worked examples of AI bias scenarios with sentence starters. Focus on Lessons 1–3. Use structured comparison tables (Western tech values vs. Māori values) before asking ākonga to write independently.
ELL / ESOL: Pre-teach key digital technology vocabulary (algorithm, bias, data, sovereignty). Allow ākonga to discuss concepts in home language before writing in English. Many ākonga from Pacific and Asian backgrounds will have rich perspectives on cultural technology tensions — create space for this.
Accelerated learners: Research a specific case of algorithmic injustice affecting Indigenous communities and write a submission-style report addressed to a government agency or tech company, using Te Tiriti obligations as the framework for critique.
Neurodiverse / Inclusion: Use accessible formats with clear headings and visual supports. Structured ethical frameworks (e.g., decision trees, weighted criteria matrices) help navigate complex AI ethics scenarios. Break the design project into weekly milestones.
Pedagogical Foundations | Ngā Tūāpou Akoranga
AI is not neutral technology — it encodes power, values, and epistemologies. Three researchers explain why AI ethics education at NCEA level requires a critical and culturally-grounded approach, not just a technical one.
→ Explore all theorists at Te Whare Ako — Teaching Theory