Māori, Pacific Peoples, ethnic communities and GenAI

In general, it’s important to protect an individual’s data and to ensure that it’s not used to create bias or harm.

This guidance is aligned with the following OECD AI principles:

Māori and GenAI

GenAI in public services presents both opportunities and challenges, particularly when involving Māori data or when it impacts Māori communities.

Māori representatives hold diverse views on Government use of GenAI systems. In particular, there are concerns among Māori, Pacific peoples and other ethnic community groups about possible discrimination resulting from the use of GenAI.

Commit to understanding Māori considerations for GenAI

We recommend understanding important contexts for Māori and the Crown. Agencies’ key considerations could include:

  • the purpose of using GenAI
  • potential impacts on Māori
  • the nature and status of Māori data involved (tapu, sensitivity and risk and noa, free from tapu)
  • Māori data governance applications.

Where Māori data is involved, we recommend aligning with existing Māori-Crown relationship approaches. Effective engagement should be simple, equitable, safe, and value-adding for Māori participants.

Managing datasets for Māori

There are many large language models that support GenAI. Given the nature and expense of these models, there is a variation – some can be narrow and/or based on other contexts not from New Zealand which do not encompass the world view or experience of indigenous groups, including Māori. Involving iwi Māori in the management and development of GenAI helps to identify potential bias and or possible discriminatory outputs.

Explore opportunities for:

  • regular engagement and, where sensible, shared decision-making with iwi Māori
  • fostering Māori-led approaches to enhance GenAI inclusivity and effectiveness in public services
  • building your team’s capability to engage confidently and respectfully with Māori.

Māori datasets GenAI Scenario

Example scenario of Māori datasets and GenAI

You’re preparing a presentation about Māori populations and practice and decide to use a public GenAI system.

You should confirm whether the data or information you need to input is considered Māori data. If it is, you should check the terms and conditions of the GenAI system, particularly privacy, data and training data sections. They and your accountable official can advise if there are any concerns, such as if data inputted into the system is shared with third parties. If you’re still unsure, you should consider not using the system.

You can talk to your manager or IT team about alternative technologies, including other types of GenAI systems, that could support your work.

Indigenous data considerations

If indigenous data is entered into public GenAI systems, there can be little control of where it ends up. This creates risk of inadvertently exposing or commodifying data without consent. Be aware that data can be considered sensitive by different people, cultures, and communities. To ensure to good data practices, we suggest the following:

  • Encourage your teams to ask ‘how should this data be treated?’ before entering it into GenAI systems.
  • Empower your teams to consider whether the information they’re inputting may put data sovereignty at risk.
  • Ensure your teams to be aware when producing content about Māori, Pacific Peoples or other ethnic communities of the perception of bias or perpetuating stereotypes.

Supporting Use of Indigenous Data and AI

The iterative learning and training capability of GenAI makes it difficult for anyone to understand how their data is being used. Extra effort should be made to ensure all involved understand the guardrails, checks and balances in place so use is safe and responsible.

Enterprise GenAI may enable closer control of data sources but requires organisations to effectively set up and manage the data the GenAI system can access. Data needs to be managed to enable the use of AI systems. Therefore, human oversight is essential for ethical, culturally aligned decisions.

Jurisdictional risks happen when data is controlled by the laws of the country where service providers store, process, or send the data. That’s why it’s important for agencies to talk to providers to know where their data is handled and understand the laws of that country.

Recommended reading

Related guidance

Utility links and page information

Was this page helpful?
Thanks, do you want to tell us more?

Do not enter personal information. All fields are optional.

Last updated