He ao anō tēnei / A different world

The systems we ship reproduce the same assumptions about what counts as value, who counts as a user, and whose knowledge counts as data. They were built into our tools long before we picked them up, and artificial intelligence is now shipping them at speed. Bias is not a bug. It is design: in schemas, in vocabularies, in models trained overwhelmingly on English. This talk reads the Māori Data Governance Model and Māori AI Governance Framework as an engineering specification most of our systems would fail, and closes with four provocations for the people who build them.

Who is this talk for?

Engineers, architects, data and platform people, and anyone who can say yes or no to an AI system. No prior knowledge of te ao Māori or Māori data sovereignty needed.

What will attendees take away?

• How to spot the assumptions already encoded in your schemas, defaults and training data.

• How to read the Māori Data Governance Model and Māori AI Governance Framework as design requirements, not compliance paperwork.

• Why where your data lives, who can compute on it, and whether Māori partners can decline, are architecture decisions rather than procurement ones.