Kia Pākiki – AI in education
Amit Sarkar and Selena Chan worked on the 2025 book Artificial Intelligence in Vocational Education and Training. They sat down with the podcast Kia Pākiki Canterbury to discuss issues around artificial intelligence (AI) as it pertains to the field of education in Aotearoa and abroad.
Kia Pākiki Canterbury – AI in education
In this Kia Pākiki Canterbury podcast, Science Communicator Tom Goulter and co-host Associate Professor Adrian Paterson from Lincoln University interview Dr Selena Chan and Dr Amit Sarkar from Ara Institute of Canterbury.
Select here to view video transcript and copyright information.
Digital technologies that talk back
Selena says that generative AI (which encompasses text platforms such as ChatGPT as well as image generators like Midjourney) represents a tipping point in the way that humans and digital technologies interact. The reason for this, she says, is that these are technologies that can talk back to us, providing feedback we may not be expecting.
AI appears to be very knowledgeable, Selena says, but sometimes the knowledge it produces may not be reliable. This presents a challenge to educators: when and how to introduce AI technology to learners. The technology can’t be allowed to replace the processes of teaching and learning, which are necessary to help ākonga become creative, critical thinkers.
Image generated by a text prompt
This image was created with a text-to-image model. The model uses written text to produce an image matching the description.
The description provided to the model was, “An oil painting of artificial intelligence thinking about itself in the style of Picasso”.
The image was generated with DALL-E2.
Selena gives the example of an exercise designed to help graphic design students identify the qualities that make a good picture. Midjourney was used to generate a large selection of pictures, and students were tasked with sifting through the results to identify the best images. An exercise like this makes full use of AI’s generative potential while keeping the tasks of thinking and discernment strictly in human hands.
There was an interesting outcome at the end of the project. “The better students had decided that AI was not for them,” says Selena.
AI to support learning, not replace it
Teaching the practical skills of a pastry cook at Ara, Selena found herself increasingly using technology to keep in touch with apprentices throughout the country. This was her entry into the e-learning space, which in turn led to mobile learning and, later, into AI.
“It’s all part of using technology to support learning, not to replace learning,” she says.
“AI literacy needs to be part of the curriculum for everyone, because AI will be pervasive and part of how society gets on with life,” says Selena. She points to a framework designed at the University of Canterbury that structures the levels at which to introduce learners to increasingly complex aspects of the technology...
“It’s not having a theoretical understanding about AI only,” says Amit. The technology needs to be incorporated into the curriculum so that students learn by doing. He uses the example of AI learning aids that interact with ākonga based on what support they need.
ChatGPT and also a lot of the others now have a study mode… So it’s more a ‘guide on the side’ than a ‘sage on the stage’.
Selena echoes this, saying AI needs to be a technology that students can call on when they need additional help with a topic or concept.
“Humans need repetition,” she says. “We need to do [something] more than once, sometimes many, many, many times.” AI can help students with the repetitive work of learning a skill until it’s been internalised.
Current issues with AI
Neither Selena nor Amit feel AI is without risks or issues. One problem with the technology as it stands is that its datasets (the vast repositories of raw data used by AI to generate responses) are based largely off text and images gathered from North American sources. Amit feels AI needs to be trained on Aotearoa users and the New Zealand Curriculum in order to be useful for education in this part of the world.
It’s really not an off-the-shelf tool at the moment, Amit says. “It can go very wrong if you think like that.”
Māori data sovereignty
Graphic illustration completed by Fuselight Creative during a presentation by Dr Te Taka Keegan titled Issues with Māori Sovereignty over Māori Language Data.
The presentation was given at the HELISET TŦE SḰÁL ‘Let the Languages Live’ – 2019 International Conference on Indigenous Languages hosted by the First Peoples’ Cultural Council and First Peoples’ Cultural Foundation in Victoria B.C., Canada.
Original illustration © Fuselight Creative.
Selena says it’s important to teach ākonga about where AI obtains its data. She also points out that this is inseparable from issues of data sovereignty, a topic of some discussion in Aotearoa. “Different cultures have different ways of looking at what happens to the information and knowledge that is part of their culture,” she says.
Another important issue is that of digital equity. Not everyone can afford to pay for the premium tiers of AI support. Selena says the COVID-19 pandemic showed educators just how many ākonga didn’t have access to laptops or unlimited data. These issues can’t be ignored when we think about how AI might change the face of education.
Educating the educators
“We can’t sit back and do nothing,” says Selena. AI is a disruptive technology that will have profound effects across society, including in education. Educators need to be having conversations about how to use the technology in the classroom and lecture hall.
What I’d love to see is that we are addressing all the problems that we thought [don’t] make sense in the common economic scenarios, but it has a valid reason to be addressed, so we become more human.
Practices and values around AI need to be created at the school level so they can be reinforced at the tertiary level, says Amit. This requires a universal framework for how we use and teach these technologies.
Selena says some questions need to be asked about AI now and in the future. Are we actually using it ethically? Is it helping our students? Are they able to do the hard yards and use AI as a supplement?
“We can’t … just use it without really thinking through the implications,” says Selena.
Listen to the full episode here, or search for Kia Pākiki Canterbury on Spotify or Apple Podcasts.
Related content
Artificial intelligence – an article about the basics of AI as it’s used today, exploring machine learning, neural networks and the face of AI in our lives.
ChatGPT – generating text and ethical concerns explores ethical, political and educational concerns surrounding the most-visible AI platform.
Bots vs Beings – the impacts of AI on life and work features panel discussions with AI experts about the uses, limitations and impacts of AI.
Useful links
Kia Pākiki Canterbury is a monthly podcast by the Canterbury branch of the Royal Society Te Apārangi. The show broadcasts on the first Monday of every month and can be found on Spotify and Apple Podcasts.
The Scaffolded AI Literacy (SAIL) Framework for Education is a university-designed resource to help prepare learners at all levels to engage constructively with artificial intelligence.
An online copy or hardcover of Artificial Intelligence in Vocational Education and Training, edited by Selena Chan, can be ordered from Springer.
Acknowledgement
This interview was excerpted from an episode of Kia Pākiki Canterbury, a monthly podcast about science, technology and humanities presented by the Canterbury branch of the Royal Society Te Apārangi.
Kia Pākiki logo
Kia Pākiki Canterbury is a monthly podcast about science, technology and the humanities presented by the Canterbury branch of the Royal Society Te Apārangi.
