ChatGPT, AI and Māori data sovereignty
Associate Professor Te Taka Keegan (Waikato-Maniapoto, Ngāti Porou, Ngāti Whakaue) is a computer scientist and Māori language advocate at the University of Waikato. He talks about an experience exploring ChatGPT with his students and discusses the implications of large language models (LLMs) like ChatGPT on the push for Māori data sovereignty.
Question for discussion
What positive applications can you imagine for AI language models to support and protect te reo Māori?
Transcript
Associate Professor Te Taka Keegan
I gave a talk to a bunch of students, Māori students from Māori schools through our Pūhoro programme recently. We had ChatGPT running and I was getting them to try it out. And I was getting them to ask some questions like “My mum keeps bugging me about doing the dishes, what can I say to her?”
So then what I got it to do was to ask it some similar questions i roto i te reo Māori, in the Māori language. And some of the language that was coming back was actually really good. Not all of these children could speak Māori, but this gave them an avenue to start learning to speak Māori.
And it was so good, it was what I term scarily good. And it’s scarily good for three main reasons. These are the three reasons I can think of for now.
The first reason is how does it know?
How does it know to speak te reo Māori? Because I’ve got a good handle on where all the Māori language databases are throughout New Zealand, and I know none of them shared them with this overseas international company. And of course the answer is, it took all of its information from public resources and social media. So everything that’s ever posted in social media – Facebook, Twitter, TikTok, YouTube, where this video is gonna be posted – it can get access to that. It pulls it all in, it learns from it. So that’s a little bit scary in terms of Māori data sovereignty.
So we’ve got this movement at the moment, Māori data sovereignty, where Māori have contro l of their own data. And this is kind of the complete opposite of Māori data sovereignty. Māori have no contro l of their own data. So that’s a concern.
The second concern is, whether we like it or not, at the moment it’s free. Eventually there will be a cost. Eventually the makers of ChatGPT will make a lot of money, and they’ll make a lot of money out of the Māori data that they’ve got. And none of that money will go back to Māori. So, so that’s a concern.
The other concern is that ChatGPT works so well and the responses are so rapid and it’s quite cheap, like it’s free – people are gonna use it. When I talked to these students, half of them were already using it. People are gonna use it.
And in terms of a Māori language activist, a lot of people are going to use it for te reo Māori. And they’re going to use it so much that, sooner or later, our language is going to start to shift. And it’ll be ChatGPT that’s controlling our language, not our traditional holders of the language.
I’m a computer scientist. I kind of like to look in the good in computer science. And I actually think there’s an opportunity for us to take this tool as it is and then repurpose it on some data that we have control of, some Māori language data that we have control of. And normally from a Māori perspective, you don’t give it Māori data, you give it iwi-related data.
So we could potentially take this tool and have it talking to us from a Maniapoto dialect or from a Ngāti Porou dialect or from a Tai Tokerau dialect. And if we fed it the right information, if we have sovereignty of our information and we can cut this tool off so it’s not sharing it with the rest of the world, then we can really start to harness the powe r of AI for Māori.
Acknowledgements
Associate Professor Te Taka Keegan, Department of Software Engineering, Co-Director of AI Institute Māori/Artificial Intelligence Institute and Associate Dean Māori for Te Wānanga Pūtaiao (Division of HECS), University of Waikato.
Video footage courtesy of the University of Waikato. The video is from a recording of the Kaupapa Kōrero panel discussion Bots vs Beings: How Will AI Impact Your Life and Work? held at the University in June 2023.
The full discussion with Professor Mike Duke, Dr Amanda Williamson, Professor Nick Agar and Associate Professor Te Taka Keegan and facilitated by Professor Bryony James can be viewed here.
