16 posts tagged with "Artificial intelligence"

The public discussion around artificial intelligence often treats American and Chinese models as if they come from completely different technological worlds. Politically and legally, they do operate under different jurisdictions. Their companies have different ownership structures, regulatory obligations, cloud environments and terms of service. Those differences matter.

But when we look at how the models themselves are trained, the broad pattern is strikingly similar. OpenAI, Anthropic, Google, Meta, Alibaba and DeepSeek all rely on some mixture of large public datasets, licensed or partner material, human-created examples, synthetic data and extensive post-training. The details vary, but the core industrial process is recognisable across borders.

For Māori organisations, this means “American model” versus “Chinese model” is not a sufficient risk framework. The better questions are what data sources were used, how transparent the developer is, what product is being used now, where current prompts are processed, what jurisdiction applies, and whether the model can be run locally without sending new Māori information back to the developer.

A common warning about generative AI says that anything typed into a chatbot becomes part of the model. That is too crude to be useful. What actually happens depends on the provider, the product, whether the account is consumer or enterprise, the settings in force, whether feedback is submitted and whether the organisation has negotiated different retention terms.

For Māori organisations and GIS teams, this matters because the difference between a public chatbot and an enterprise or local system can be substantial. A planning report, shapefile attribute table, field note, Treaty research document or set of candidate place names may be handled very differently depending on the service used.

The first technical point is simple: entering a prompt normally causes inference. The already-trained model processes the current context and produces a response. The model is not usually changing its weights while the user types. After that inference, however, the service may retain the interaction, and in some consumer products the retained material may be eligible for later model improvement. Those are separate stages.

This article looks at current published policies as at 4 September 2026. They change, so the exact product terms should always be checked before important Māori or organisational data are used.

Local AI has become much more useful in 2026. A normal desktop computer can now run a capable language model, search a collection of PDFs, extract structured information, help write Python or QGIS expressions and keep the working material on the machine. That makes it particularly interesting for Māori GIS, where the question is often not only whether AI can do the work, but where the information goes while it is doing it.

The most useful local workflow is not a private version of ChatGPT that somehow knows everything about Māori GIS. It is more modest and, for research, more useful. Give the system documents you are allowed to work with. Ask it to find candidate information. Keep the original wording and source reference. Have a person review the result. Check any geographic match against an appropriate source. Only then let the reviewed information become GIS data.

That distinction matters. A language model is quite capable of reading a reference to an old kāinga, deciding what modern place it probably means and supplying coordinates that look perfectly respectable. A GIS will then plot those coordinates without showing the slightest concern about whether they were invented. Local AI changes where the mistake was made. It does not make the mistake authoritative.

The practical question is therefore: what can we now run locally, what is genuinely offline, what appears promising for te reo Māori, and where should AI stop before the map begins?

A few years ago the AI part of a GIS workflow was simple enough. Open ChatGPT, type a question, see what comes back. Sometimes it was useful. Sometimes it confidently invented a Python function. At least the administrative overhead was low.

That has changed. A Māori GIS practitioner wanting help with an ordinary piece of work can now face a small set of decisions before the work starts. ChatGPT, Claude, Gemini, Copilot or something more specialised? A fast model or the one that thinks for longer? Ordinary chat, a project, a research mode, a coding agent or a persistent workspace? Is the task worth using a limited premium allowance? Does the system need three source documents or the whole project folder? Can it use the web? Can it run code? Will it remember anything later? Most importantly for Māori GIS, what information are we about to give it, and should that information leave the environment it is already in?

This is starting to look like a professional skill in its own right. Knowing AI no longer means being good at typing questions into one chatbot. It increasingly means knowing how to select and manage several different kinds of machine assistance, give them the right amount of context, use expensive capability where it matters, keep data under appropriate control, and check the result before it quietly becomes part of the GIS.

One of the first things GIS teaches you is confidence. A line appears on the screen cleanly, precisely and apparently without argument. You can zoom into it, measure it, colour it, calculate its area and print it at A0. Once information has travelled through a computer, it acquires a peculiar authority. It looks as though somebody, somewhere, must know exactly where it is and exactly what it means. One of the first things Māori claimant mapping taught me was almost the opposite. Historical land blocks did not always line up neatly with the modern cadastre. Old plans could disagree with one another. Rivers moved. Survey descriptions depended on natural features that had changed or disappeared. Names varied between sources. Areas written on historical records did not always agree with areas calculated by modern GIS. Sometimes a plan was missing altogether, and sometimes two apparently authoritative documents contradicted one another.

In one of the Northland mapping methodologies we had a wonderfully unfashionable description for this: State of Unknown. Rather than forcing the available information to produce an answer simply because the software was capable of drawing one, we recorded that the answer remained unresolved. At the time it was a practical classification for awkward historical geography. Over the next twenty years it became something closer to a professional philosophy. The temptation in GIS is always to resolve, align, categorise and complete. State of Unknown was a reminder that sometimes the more professional thing is to preserve the uncertainty.

What changed more slowly for me was the way I thought about the thing underneath the map. In conventional GIS the land can easily become a surface on which information is located. Parcels sit on it, roads cross it, rivers run through it and points are placed where events occurred. Working alongside Māori claimants, practitioners, researchers, iwi and hapū repeatedly challenged that habit. Whenua was not simply the backdrop to the information. It sat within relationships, whakapapa, occupation, memory, responsibility, history and identity. The map could help represent parts of those relationships, but it could not contain them all.

I am not Māori, and I do not claim to explain te ao Māori or speak for the people whose knowledge shaped the projects I worked on. My part in this history has been much more practical. I have built maps, recovered old plans, configured GIS systems, organised workshops, trained people, supported conferences, found data, developed resources, fixed broken things, produced programmes, dealt with government agencies and occasionally discovered that an ambitious twenty-first-century mapping project was being held together by a cardboard box of DVDs and somebody remembering where the extension lead had gone. What changed my practice was being in rooms where somebody else understood the place, history or meaning far better than I did and learning that the technical role was most useful when it supported that authority rather than competing with it.

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.