Guidelines for the best-practice use of generative artificial intelligence in research in Aotearoa New Zealand

I am pleased to see Te Apārangi taking leadership to establish Guidelines for the best-practice use of generative artificial intelligence in research in Aotearoa New Zealand and I acknowledge and thank those who put the time, energy, and expertise into developing them. I appreciate the care taken and coverage, and although I raise critical questions here, I commend the general position expressed. 

The guidelines rightly place questions and obligations in relation to te Tiriti, mātauranga Māori, Māori, Pacific and Indigenous data sovereignty, and peoples, etc at the centre of the discussion. Obligations under te Tiriti are also connected to the briefly noted environmental impacts of GenAI (related to the responsibility to protect te taiao). But I am disappointed at how the discussion of responsible use is not (also) founded on a base of ethics and morals that centres the environmental impacts as fundamental. And, therefore, which would place environmental impact as foundational in questions of when, how, but more importantly if, we should be using GenAI in our research. Too often, discussions of environmental impact of GenAI are treated as a ‘but of course there are environmental impacts we should consider’. Such acknowledgement, in discussion and policy which then doesn’t treat them as primary, effectively makes them a discursive disclaimer, and allows for them to be disregarded. Something we move on from without having to deal with.  

Imagine how different our questions about the ethical, moral and responsible use of GenAI in research (and higher education) might look if we had to justify our choices in environmental impact terms (rather than, say, efficiency terms, which centralise the neoliberal values of productivity which have been so harmful to the research and education sectors). Maybe our question should start with a something like “what possible justification can I have for deploying this new technology, if I can do just as well without it?” In some cases, GenAI may indeed offer benefits that far outweigh any costs – and offer tools that go well beyond what we could do before. But in many others, it may more simply be that novelty, promises of efficiency, or concerns about being “left behind” are the real drivers for use (I touched on some of these concerns in a recent keynote I did on GenAI and qualitative research). To me, none of these justify something we know not only to be environmentally harmful, but which – in most cases – is also trained and developed on stolen materials, for the financial benefit of already very wealthy people.  

Ngā mihi nui

Professor Virginia Braun FRSNZ 
School of Psychology Te Kura Mātai Hinengaro 
Faculty of Science Te Whare Pūtaiao
Waipapa Taumata Rau – The University of Auckland