Useful video: safe and sovereign AI for Māori GIS

The May 2026 Ngā Poutama Matawhenua wānanga, AI Haumaru, AI Motuhake mō te Māori GIS kaupapa, focused on practical uses of artificial intelligence while protecting Māori GIS data and maintaining control over sensitive information.

The recording and related material are collected on the Ngā Poutama Matawhenua resources page, where the current links can be maintained in one place.

The session was held on Friday 22 May 2026 as an extended online wānanga. The framing matters. AI haumaru is not only a cyber-security question, and AI motuhake is not simply a technical setting. For Māori GIS, the deeper issue is whether people retain practical authority over what information is supplied, what the system is allowed to do with it, how outputs are checked and whether the technology strengthens or weakens rangatiratanga over the mahi.

Why it is useful

AI tools can help with repetitive GIS and information work, but they can also create new ways for information to leave organisational control. That matters when a project includes culturally sensitive locations, whenua information, unpublished kōrero, personal information or other restricted material.

The session looks at AI in a Māori GIS context rather than treating it as a generic productivity tool. Practical uses include helping structure or clean non-sensitive tabular data, drafting metadata and plain-language descriptions, creating formulas, scripts and processing ideas, helping interpret software errors, summarising public source material and using local desktop AI where information should not be sent to an external service.

AI can also lower technical barriers. A person who is not a programmer may be able to understand a script, diagnose an error or build a simple workflow with assistance. That can support capability and mana motuhake if it helps people do more of their own mahi. The same tool can create a new dependency if nobody understands or can verify what the model produced.

Keep the data decision separate from the AI decision

Before putting information into an AI system, first decide whether that information is allowed to leave the system or device where it is currently held. Useful questions include whether the information is already public, who has authority to decide how it is reused, whether the service retains prompts or uploaded files, whether material is used to improve models, whether the same task can be done locally, and whether a generalised or synthetic example would be enough.

AI may be useful without ever receiving the sensitive source data. That is an important shift in mindset. The question should not be “How can we get all of our information into the AI?” but “What is the minimum information the tool actually needs for this task?”

That leaves more room for rangatiratanga. The people responsible for the information decide what crosses the boundary into the external system, rather than allowing the convenience of the tool to make that decision by default.

AI is not cultural authority

Generative AI can write very confidently about subjects it does not genuinely understand. It may produce a plausible explanation of an ingoa wāhi, iwi history, customary relationship or boundary based on fragments of material it has encountered online. Fluent language can make that answer look much stronger than its evidence.

For Māori GIS this requires a clear boundary. AI can assist with technical work, public-source research and drafting, but it does not acquire authority over Māori knowledge because it can generate a convincing paragraph about it. Where a question depends on whakapapa, local history, tikanga or knowledge held by particular people, the model should not be allowed to substitute generated text for the people who hold that authority.

This is one place where the old GIS discipline of provenance becomes even more important. Ask where the answer came from, what can be verified and whether the system is distinguishing evidence from inference. If it cannot, treat the output as something to check rather than something to believe.

Kaitiakitanga and digital responsibility

Holding information creates responsibilities that are not solved by a privacy checkbox. Kaitiakitanga in this context is larger than conventional data stewardship. It includes thinking about whether the information should be put into the tool at all, what future copies may exist, who could gain access, whether the output exposes relationships that were not previously public and what responsibility is owed to the people represented in the data.

This also means that deletion, withholding and local processing can be positive design choices. More data in the model is not automatically better Māori GIS.

Related 2026 resources

Te Pā Tūwatawata provides a useful current example of Māori-owned, iwi-designed digital storage and the wider question of where Māori information is held.

Te Mana Raraunga remains an important source for Māori data-sovereignty principles and current policy submissions.

The University of Waikato's Indigenous Data in Artificial Intelligence research also provides useful background on Māori algorithmic sovereignty and Indigenous approaches to AI.

A practical principle

The most useful AI workflow is often the one that leaves the sensitive information where it is. Public datasets, sample files, synthetic records and de-identified examples can be enough to develop a schema, write a QGIS expression, create a script or test an approach. The final work can then be applied locally under the authority of the people responsible for the real information.

That is consistent with the wider Ngā Poutama principle that technical learning should increase capability without requiring participants to surrender their data. AI should be another tool available to the kaupapa, not a new reason for the kaupapa to surrender control to the tool.

For the rest of the 2026 wānanga and related mapping resources, see Ngā Poutama Matawhenua videos and StoryMaps.