❓ Question 1: What should be in an NZ School AI Use Policy?

Dr Craig’s Answer: A secure, robust board policy must go beyond legacy IT guidelines and define four non-negotiable boundaries:

• Data Privacy & Sovereignty: Explicitly forbidding the uploading of sensitive student databases, PAT results, or report comments into public commercial AI spaces.
• Institutional Sandboxes: Mandating that school credentials utilize corporate-grade sandboxes where data is strictly isolated and never ingested for public model training.
• Academic & Assessment Integrity: Aligning rules to support NCEA and NZQA guidelines, ensuring student “thought scaffolding” is protected and oral defenses are incorporated.
• Linguistic & Cultural Respect: Specifying how RAG tools can safely generate localized learning cards using accurate Te Reo headings and culturally responsive parameters.

❓ Question 2: How can AI safely reduce teacher planning time without hurting quality?

Dr Craig’s Answer: Rather than using standard, generic chat systems, school leaders should upskill teachers to build custom, structured “Gems” or “GPTs”. By copying and pasting direct language from the Ministry of Education’s refreshed curriculum milestones (such as Year 4 Geometry progress descriptors) into the Gem’s private knowledge context, the assistant automatically drafts lesson sequences, rubrics, and inclusive resources in minutes with zero risk of “hallucinations.”

❓ Question 3: How do we prevent our school from generating lazy “workslop”?

Dr Craig’s Answer: “Workslop” refers to AI-generated materials that look polished and articulate on the outside but lack structural value on the inside. DPs and principals report wasting thousands of hours re-writing lazy, unverified AI outputs. Your board policy must establish a strict “verify-first” culture: and ensure that any parent newsletters, student guides, or board documentation undergo strict human curation before publication.