🔍 THE BOTTOM LINE
AI in education is past the “should we?” phase. The question now is: are we building it in a way that helps teachers, or are we building it in a way that looks good in policy documents?
1. ✅ NZQA’s AI Grading: 3.5 Weeks Faster Results
The story: NZQA used AI for marking writing literacy assessments in 2025, and results went back to students 3.5 weeks earlier than the previous year. This is the kind of concrete benefit that cuts through the hype.
Key facts:
- AI used for writing literacy assessment marking in 2025
- Results returned 3.5 weeks earlier than 2024
- No reported accuracy issues (yet — full evaluation pending)
- Part of broader AI-in-assessment trial
Why it matters: Turnaround time isn’t just convenience — it’s pedagogy. Feedback that arrives 3 weeks earlier is feedback students can actually use. This is AI solving a real problem, not creating a new one.
Our take: This is the template for “good AI in education”: specific use case, measurable improvement, human oversight retained. More of this, less “AI will transform learning” vagueness.
2. ⚠️ NZ’s AI Framework for Schools: Gender-Blind by Design
The story: Jo Cribb (ex-Ministry for Women CEO) analysed NZ’s Public Service AI Framework — the word “gender” appears zero times. For a framework governing AI that will affect students, teachers, and school operations, that’s not neutrality. It’s blindness.
Key facts:
- Public Service AI Framework (2025) doesn’t mention “gender” or “women”
- Same framework guides school AI use
- Women concentrated in administrative/clerical roles most exposed to AI disruption
- Girls face different AI risks (deepfakes, harassment) than boys
Why it matters: If your AI framework doesn’t account for how technology affects different groups, you’re not being neutral — you’re ignoring evidence. Girls and women face different AI risks. A framework that doesn’t see that won’t protect them.
Our take: This isn’t woke box-ticking. It’s risk management. If you’re writing AI policy for schools and haven’t asked “how does this affect girls differently?”, you haven’t finished the assignment.
3. 📚 OECD Digital Education Outlook 2026: What’s Actually Promising
The story: OECD released its 2026 Digital Education Outlook, evaluating AI tools by evidence, not marketing. The findings: most “AI education” products are still solution-looking-for-problem.
Key findings:
- Most AI education tools lack independent validation
- Promise areas: adaptive tutoring, automated admin, accessibility support
- Risk areas: surveillance, bias in assessment, teacher displacement rhetoric
- OECD emphasizes “AI for teachers” not “AI instead of teachers”
Why it matters: The OECD is the closest thing we have to an international education standards body. Their “show me evidence” approach is the antidote to “trust us, it’s AI” marketing.
Our take: If an AI education product can’t point to independent validation, it’s a pilot project — not a procurement decision. NZ schools should adopt this standard.
4. 🇳🇿 Practical AI Guide for NZ Teachers (2026 Edition)
The story: GenAITraining.co.nz and EdTech NZ both updated their practical AI guides for NZ teachers. The tone shift is notable: less “here’s what AI can do,” more “here’s what you can do tomorrow.”
Key resources:
- GenAITraining: “AI for NZ Teachers and Schools: A Practical Guide for 2026”
- EdTech NZ: Teacher voice on AI framework needs
- Ministry of Education: Generative AI guidance (updated Nov 2024, due for refresh)
- Focus: curriculum-connected, practical, appropriate for teachers (not tech people)
Why it matters: Teachers don’t need another “AI is coming” presentation. They need: lesson plans that work, assessment strategies that don’t break, and policies that don’t assume the worst of students.
Our take: The best AI professional learning for teachers is: practical, curriculum-connected, and respects teacher expertise. If it talks down to teachers or assumes they’re technophobes, it’s already failed.
5. 🌍 Global Trends: 7 AI Uses Transforming Schools (2026)
The story: Third Space Learning analysed how schools are actually using AI in 2026. The shift: from experimentation to practical implementation. Focus on tools that save teacher time, protect budgets, and improve student learning.
Seven practical uses:
- Automated admin — Attendance, scheduling, parent communications
- Differentiated practice — Adaptive exercises for students who need extra support
- Accessibility — Text-to-speech, translation, readability adjustment
- Teacher planning — Lesson plan generation (with teacher review)
- Feedback support — Draft feedback that teachers refine
- Resource creation — Worksheets, quizzes, examples (teacher-curated)
- Professional learning — AI as coaching tool for teachers
Why it matters: These aren’t transformative in the “AI will replace teachers” sense. They’re transformative in the “this saves 5 hours a week” sense. That’s the kind of transformation teachers actually want.
Our take: Boring is good. Boring means AI is becoming infrastructure, not a demo. The schools winning with AI are the ones using it to give teachers time back, not to replace them.
6. 🎓 What’s Missing: NZ Teacher Voices in AI Policy
The story: Every AI-in-education document references “teachers” — but how many were written with actual teacher input? The gap: NZ classroom teachers aren’t in the room when AI policy is written.
The evidence:
- Ministry guidance written by policy teams, not practitioners
- School leaders consulted, but classroom teachers rarely
- AI vendors present at every table
- Result: policy that sounds good, breaks in practice
Why it matters: Teachers know what breaks. They know what saves time and what creates more work. If you’re writing AI policy without a teacher who’s used AI in their classroom this term, you’re writing fiction.
Our take: NZ needs a teacher advisory group for AI in education. Not “consultation” — actual advisory power. The people implementing the policy should help write it.
🔍 THE BOTTOM LINE (reprise): AI in education is past the hype phase. NZQA’s results prove it can work. The gender-blind framework proves we’re still getting governance wrong. Teachers need practical tools, not policy poetry. And if we’re not listening to classroom teachers, we’re not building for classrooms.
Sources:
- School News NZ — “Can AI accurately help teachers’ grading turnaround time?” (7 May 2026)
- Newsroom — “Immediate action needed to stop AI avalanche burying election debate” (Jo Cribb, 6 May 2026)
- OECD — “Digital Education Outlook 2026” (2026)
- GenAITraining.co.nz — “AI for NZ Teachers and Schools: A Practical Guide for 2026”
- EdTech NZ — Teacher voice on AI framework (2024-2026)
- Ministry of Education — “Generative AI” guidance (updated Nov 2024)
- Third Space Learning — “How schools are using AI in 2026: 7 Practical Use Cases” (1 Apr 2026)