Te Kete o Karaitiana Taiuru (Blog)

A Māori cultural perspective of AI/machine sentience

This is a copy of the opinion piece I wrote in the publication Goffi E. R., Momcilovic A., et al. (Eds). Can an AI be sentient? Multiple perspectives on sentience and on the potential ethical implications of the rise of sentient AI. Notes n° 2, (2022). Global AI Ethics Institute.

The New Zealand Supreme Court decision on the Ellis case had not been published at the time of writing this opinion piece, therefore it is not included. Had it been, I would have included a section of the tikanga in common law. However, I will write a separate article on that.

Based on the interview with Google’s LaMDA ; I discuss the implications to Māori if an AI sentient identifies as Māori or identifies as non-Māori New Zealander and the risks of colonisation by the sentient AI. I then consider the Māori ethical considerations and impacts to Māori culture.

Sentient AI self identifies as Māori?

From a government perspective, the definition of who can be Māori is defined in numerous pieces of New Zealand Legislation such as the Treaty of Waitangi Act 1975 and the Electoral Act 1993 as “’Māori’ means a person of the Māori race of New Zealand; and includes any descendant of such a person”.

Unique to the rest of the world, the New Zealand Government has granted legal personality to two mountains and one river due to their association with Māori tribes. This means those natural features have the same legal status as an individual person.

In 2014, legal personality as granted to Te Urewera – the mountainous region bordering Hawkes Bay and the Bay of Plenty . In March 2017, the Whanganui River received the status of a legal person and then later in 2017, Taranaki iwi signed a Record of Understanding to state their shared intention that legal personality will be granted to Taranaki Maunga (Mount Taranaki) as well .

Those two Acts of 2014 and 2017 and the Record of Understanding give all the rights, powers, duties, and liabilities of a legal person to these natural features based on the ontological understanding that the features have as living and as the spiritual ancestors to Māori and Māori tribes.

From Māori perspective, anyone is Māori who has an ancestry to Māori person, deities, and the environment. Any natural object is considered to be a Taonga (precious object of Māori heritage) if it has a genealogical connection to a Māori deity. In 2021, the statutory Waitangi Tribunal heard a claim that Māori Data is a Taonga in the Wai-2522 claim The Trans-Pacific Partnership Agreement (TPPA). The Tribunal agreed with Māori claimants. The flow on impact from this decision is that any AI that uses Māori Data is also a Taonga. If the AI is built using Māori Data or by developers of Māori descent, the AI sentient could claim to be a Māori.

Sentient AI self identifies as non Māori?

In New Zealand, non-Māori have many Māori terms including the common word Pākehā which is historical, does not always include all non-Māori and often has controversial connotations . The recent adaption of the term Tangata Tiriti describes the commitment non-Māori have to recognise Te Tiriti o Waitangi (A founding documents of New Zealand) and He Whakaputanga (A founding constitutional document) to build a relationships with Māori, to understand the colonial history of New Zealand and to commit to the continuing fight for Māori self-sovereignty. It also acknowledges that New Zealand is a multicultural country with all races brought together with Māori under the Treaty.

If the AI identifies as being sentient, then it should be given data about Te Tiriti and He Whakaputanga as well as the United Nations Declaration of the Rights of Indigenous Peoples so it can understand the role of Tangata Tiriti and not become bias against Māori.

Māori Ethics test

To explore if an AI sentient is ethical to Māori, I will use the Māori cultural ethics test designed to discuss ethically controversial issues.

Test 1: the tapu test

All Māori people, species and physical things have a decent to Māori deities, usually Ranginui and Papatūānuku. The AI is tapu (sacred) as anything in Māori is sacred, including Māori Data from the Māori deities Tāne Mahuta and Rehua. As AI can aid in decision making, it has another primary Māori deity Hinengaro the deity of thoughts, conscious, instinct etc.

The AI sentinel passes this test as these deities do not conflict with each other.

Test 2: the mauri aspect

Every living thing and physical object has a mauri (a life force). The sentient AI will have a mauri of the developers and from the Māori Data it uses. This means that the AI will need traditional and Māori ethical considerations. To protect the mauri of the people involved and the AI, data from the constitutional documents and a group of learned individuals for cultural advise.

Test 3: the take-utu-ea or TUE test

This opinion piece and test has highlighted a way to avoid breaching Māori ethics, therefore this test is not applicable.

Test 4: the precedent aspect

There is traditional Māori knowledge that is a precedent for a Māori sentient AI. There is a myriad of stories about ancestors changing themselves into various objects and other species and the transfer of knowledge.
Māori still listen and watch the environment to make decisions about harvesting, fishing, and other daily tasks. The myriad of deities are still worshiped to assist with decision making. A sentinel AI is merely a modern form of a decision maker.

The fact that two mountains and one river have been given legal personhood in New Zealand by the government, also establishes a precedent that a sentinel AI could be granted personhood for the same reasons as the mountains and river’s connection to Māori tribes where they are geographically located.

Test 5: the principles aspect

It is essential that any AI is trained using Māori ethics, New Zealand’s constitutional documents and related data to ensure that it is not bias against Māori.

Test 5.1: whanaungatanga
To pass this test, the families of the knowledge providers in the data and the Māori developers should be treated as a family group and support each other and to interact with the sentient AI.

Test 5.2 Manaakitanga
Results from the previous tests reflect that the sentient AI should be treated as a person.

Test 5.3: Mana
It is possible that the sentient AI could revive knowledge that has been lost due to colonisation and this should be anticipated in advance.

Test 5.4 noa
This is the principle of acceptance by Māori society. Using a mixture of traditional knowledge outlined in this test and discussions how the sentient AI can remove human bias from data sets and assist Māori, the state of noa (normality and acceptance) will be reached by the majority.

Test 5.5: tika (Conclusion)
This principle seeks consideration of the previous tests and asks if a sentient AI is acceptable to Māori and the public.

A Sentient AI could use legislation, legal precedents, and traditional Māori knowledge to state it is a Māori. but his would create a number of traditional issues such as family members, tribal affiliations, access to land and other natural resources, succession planning etc, that would need to be discussed further at tribal and a family level. The legal personhood status of the mountains and river could be used for guidance.

A sentient AI could benefit Māori Peoples if appropriate Māori ethics are applied, including data about New Zealand’s constitutional documents.

DISCLAIMER: This post is the personal opinion of Dr Karaitiana Taiuru and is not reflective of the opinions of any organisation that Dr Karaitiana Taiuru is a member of or associates with, unless explicitly stated otherwise.

One response to “A Māori cultural perspective of AI/machine sentience”

  1. […] In my published chapter in Goffi E. R., Momcilovic A., et al. (Eds). Can an AI be sentient? Multiple perspectives on sentience and on the potential ethical implications of the rise of sentient AI. Notes n° 2, (2022). Global AI Ethics Institute., available as a blog post. […]

Leave a Reply

Archive