New Zealand's education sector spent most of 2024 arguing about whether students are using ChatGPT to cheat. Valid concern. Wrong focus. While we debated plagiarism detection, the sector missed the more significant question: how does AI transform learning, teaching, and the administration that drains every educator's energy?
I've spent the last year talking to educators, administrators, and policy people across the NZ education sector. The pattern is consistent. At every level - primary, secondary, tertiary - the conversation starts with concerns and ends with unrealised potential. It's time to flip that.
The State of Play
Let's be honest about where NZ education sits with AI at the end of 2024:
Policy is catching up but fragmented. The Ministry of Education has issued guidance. Individual institutions have developed their own policies. There is no unified national framework for AI in education. Every school, polytechnic, and university is making it up as they go.
Student AI use is pervasive and ungoverned. Students are using AI tools extensively, regardless of institutional policy. Banning ChatGPT in 2024 is like banning Wikipedia in 2008 - technically possible, practically pointless, and educationally counterproductive.
Educator AI use is nascent. Some teachers are experimenting with AI for lesson planning, content creation, and assessment design. Most aren't. The barriers are time, training, and institutional support - not technology.
Administrative AI adoption is nearly zero. This is the biggest missed opportunity. Education administration is drowning in manual processes that AI could handle effectively today. Timetabling, reporting, compliance documentation, communications - all ripe for AI assistance.
73%
of NZ tertiary students report using AI tools for study-related activities
Source: NZUSA, Student Technology Survey, 2024
NZ Education AI Opportunities: Potential vs Readiness
Source: RIVER Group analysis, 2024
The Three Opportunities
1. Personalised Learning (High Potential, Medium Readiness)
Every educator knows that students learn at different paces, in different ways, with different strengths. Every educator also knows that class sizes, time constraints, and workload make genuine personalisation nearly impossible with current approaches.
AI changes this equation. Not by replacing teachers, but by extending their reach.
What's possible today:
- Adaptive practice. AI that adjusts question difficulty and topic focus based on student performance. Not a new concept - adaptive learning platforms have existed for years. But GPT-4-era AI makes the adaptation more nuanced and the feedback more natural.
- Personalised feedback. AI that provides detailed, specific feedback on student work. Not grading - feedback. "Your argument would be stronger if you addressed the counterpoint in paragraph three" rather than "B+." The AI handles the volume. The teacher handles the judgement calls.
- Differentiated content. AI that presents the same concept at different levels of complexity, in different formats, with different examples. A struggling student gets a simpler explanation with concrete examples. An advanced student gets a deeper exploration with connected concepts.
What's not ready:
- Fully autonomous tutoring that replaces teacher interaction. The technology isn't reliable enough, and the pedagogical relationship between teacher and student can't be replaced by technology.
- Assessment of complex skills (critical thinking, creativity, collaboration). AI can assess factual knowledge and procedural skill. Higher-order skills require human judgement.
2. Administrative Automation (High Potential, High Readiness)
This is where I'd start if I were running a school or university right now. The administrative burden on NZ educators is enormous and growing. Compliance reporting, timetabling, communications, documentation, enrolment processing, special needs coordination - the list is endless.
What's possible today:
- Compliance documentation. AI that drafts compliance reports from existing data. The education sector generates enormous amounts of reporting for ERO, the Ministry, TEC, and NZQA. Much of this is pulling data from multiple systems and formatting it. AI handles this well.
- Communication drafting. Parent communications, student notifications, inter-agency correspondence - AI can draft these from templates and context, with human review before sending.
- Timetabling optimisation. AI that generates and optimises timetables considering constraints (room availability, teacher schedules, student course combinations). This is a well-understood optimisation problem that AI handles efficiently.
- Meeting summarisation. Board meetings, staff meetings, IEP meetings - AI can transcribe, summarise, and extract action items. A significant time saving for administrators.
Education is debating whether AI will replace teachers, while teachers spend 30-40% of their time on admin AI could handle today. The AI isn't coming for teaching jobs - it's coming for the paperwork that stops teachers from teaching.
Tim Hatherley-Greene
Chief Operating Officer
3. Assessment Innovation (Medium Potential, Low Readiness)
This is the most contentious area and the one with the most transformative potential. AI doesn't just challenge how we detect cheating. It challenges what assessment means.
The rethink:
If AI can write an essay, the essay isn't testing what we thought it was testing. It was testing recall, synthesis, and written expression. AI can do all three. So what are we actually trying to assess?
- Can the student identify the right question to ask?
- Can they evaluate whether the AI's output is correct?
- Can they improve, critique, or extend the AI's work?
- Can they apply knowledge in contexts the AI hasn't seen?
- Can they collaborate, present, and defend ideas?
These are more valuable skills than essay writing. They're also harder to assess. The opportunity isn't to AI-proof our existing assessments. It's to redesign assessment for an AI-augmented world.
What's emerging:
- Process-based assessment (evaluating how students work, not just what they produce)
- AI-collaborative assessment (students use AI as a tool and are assessed on how effectively they use it)
- Oral and presentation-based assessment (harder to outsource to AI)
- Portfolio-based assessment (cumulative evidence of capability over time)
NZQA is beginning to engage with these questions. But institutional change is slow, and the assessment infrastructure (NCEA, university grading systems) is built around formats that AI challenges.
The Equity Dimension
Any discussion of AI in NZ education must address equity. Not as an afterthought. As a primary concern.
Digital divide. Students with personal devices and home internet access have unlimited AI access. Students without have little or none. AI in education risks widening an already significant equity gap.
Cultural relevance. AI models trained predominantly on English-language, Western educational content may not serve Māori and Pacific students well. Culturally responsive education requires culturally informed AI tools, and those largely don't exist yet.
Rural access. NZ's rural schools face connectivity constraints that urban schools don't. Cloud-dependent AI tools assume reliable internet access.
Teacher capacity. Decile 1 schools typically have less capacity for technology adoption (fewer support staff, less professional development budget, higher staff turnover). The schools that most need AI's efficiency gains are least resourced to adopt it.
Actionable Takeaways
- Start with administrative AI. Lowest risk, clearest time savings, no student-facing concerns. Free up teacher time for teaching.
- Rethink assessment before AI-proofing it. The plagiarism detection arms race is unwinnable. Redesign assessment for an AI-augmented world instead.
- Invest in educator AI literacy. Not one-off workshops. Sustained professional development that helps teachers understand, evaluate, and use AI tools effectively.
- Address equity first. Before deploying AI tools that require devices and connectivity, ensure equitable access. Otherwise AI becomes another vector for educational inequality.
- Engage students as partners. Students are already using AI. Bring their experience into the conversation rather than policing it. Their practical knowledge of AI tools often exceeds their teachers'.
- Demand NZ-developed educational AI. International educational AI tools don't account for NZ curriculum, te reo Māori, NZ English, or our assessment frameworks. Support local development.