Further reading
The course stands on a body of published work. These are the pieces that support its narrative — our own essays and papers, the companion Governance Course, and the external sources the argument draws on. Read them to go deeper on any module; none is required to complete the course.
The companion course
- Reference glossary — 296 plain-language terms for sovereign, governed AI, EN + te reo Māori.
- Village Governance Course — the other half of the discipline: the record that can be relied upon, the constitution, and the decision a board can demonstrate afterwards. Start with its 5-minute primer.
Our essays & papers — agenticgovernance.digital
- Taonga in the Latent Space — a community-scale world model; why an agent must foresee what its action will do. (Modules 1, 6)
- AI that stays in its lane — bounded competence and built-in deference. (Modules 1, 5)
- An alternative to Big Tech AI — the sovereign path, and why it is durable. (Modules 1, 6)
- Plural values, living organisations and AI — held in kōrero, not collapsed to a number. (Module 3)
- Who authors, who governs — the people remain the authors; the machine serves. (Modules 3, 6)
- The off switch is not enough — why command must be structural, not a button. (Module 5)
- A control tower, not a watchtower — the shape of oversight that holds. (Module 5)
- Tamper-evident governance — the record that can be relied upon afterwards. (Module 5)
- Federate, don't align — many small, sovereign models composing, not one monolith. (Module 6)
- Sovereignty without dominance — sovereignty that does not become another centre. (Module 6)
- A Village you can run yourself — what the sovereign platform under all this actually is.
- Your model, your walls — isolation and ownership of the record.
- For the technical foundations: Sovereign-Record Architecture and the Aotearoa NZ Agentic-AI Framework.
- More across the blog and field notes.
External sources the argument draws on
- Yann LeCun — A Path Towards Autonomous Machine Intelligence — the world-model architecture behind Module 1.
- V-JEPA 2 (Meta AI) — a world model trained on video; a working example of prediction in representation space.
- AI New Zealand / NewZealand.AI — the practical-adoption community this course is pitched alongside.
- Fast Forward Aotearoa — Ben Reid — a national technology strategy for New Zealand, sovereignty-first.
Foundations — te ao Māori
The wider programme draws on, and defers to, work that belongs to tangata whenua: Te Kāhui Raraunga's Māori Data Governance Model, and Dr Karaitiana Taiuru's Kaupapa Māori AI Framework — He Tangata, He Karetao, He Ātārangi. Full citations are given in the sources of Taonga in the Latent Space. What counts as appropriate use in te ao Māori contexts is for tangata whenua to determine.