A New Zealand health professor says there is still significant work needed to ensure AI systems perform well for Māori and Pacific people.
Professor Robyn Whittaker, co-director of the University of Auckland’s TRANSFORM Research Centre, told the Te Poutoko Ora a Kiwa Research Symposium that many AI systems used globally are trained on datasets that do not adequately represent Māori and Pacific populations.
“It is very unlikely these tools have been properly tested on Māori and Pacific communities,” she said.
“There is still significant work needed to ensure they perform well for our people.”
She said her team has been working on evaluating AI tools through an advisory group, with a focus on safety, ethics, equity, and Pacific perspectives.
She said predictive algorithms were used during the COVID-19 pandemic to identify Pacific patients most at risk of hospitalisation, but tools like that need high-quality, locally relevant data.
“We need to work together to ensure AI is implemented in ways that are safe, ethical, and equitable for all,” she said.
“We are particularly concerned about bias, data sovereignty, and the control of indigenous data.”
Professor Robyn Whittaker, co-director of the University of Auckland’s TRANSFORM Research Centre, told the Te Poutoko Ora a Kiwa Research Symposium that many AI systems used globally are trained on datasets that do not adequately represent Māori and Pacific populations.
University of Auckland
Co-director of Te Poutoko Ora a Kiwa, Professor Sir Collin Tukuitonga, said new technologies should not widen existing inequities, but support better, fairer health outcomes across the Pacific.
A report delivered to New Zealand’s government before the last election in 2023 noted health inequities affecting Māori and Pasifika, including in relation to AI.
It said New Zealand’s health system had “systemically underperformed” in diagnosing and treating Māori, Pasifika, and women, and that “human fallibility… can lead to unequal outcomes”.
It said AI tools may be better than clinicians at “flagging rare conditions that disproportionately affect disadvantaged groups”, “assuming AI tools were developed and fine-tuned on diverse data that reflected differences in disease prevalence between groups”.
The report, from the Office of the Prime Minister’s Chief Science Advisor (OPMCSA), stated poor data had contributed to health inequalities.
“A recent report on health equity for Pacific people pointed to significant data challenges for this group, including disputed ownership of primary care data broken down by ethnicity and no requirements around secondary care data, despite national policy to promote health equity for Pacific.
“AI tools trained on data which poorly capture ethnicity will perform less well for some ethnic groups than others overall, while individual patients could be disadvantaged if ethnicity is incorrectly entered into an AI tool used in their care.
“For Pacific peoples, understandings of health are holistic, taking into consideration physical, mental, spiritual, social, and economic wellbeing of the community.
“Despite the increasing diversity, several cultural values are consistent across Pacific groups. The appropriate respect and consideration for these values as AI is implemented into our healthcare delivery services will form a strong foundation for a thriving Pacific population in Aotearoa.”
More AI – digital health leaders hear about new opportunities
Nine To Noon
Since then, AI’s place in society has only become more pronounced.
New Zealand doctors said the new AI scribe in EDs was saving them up to 10 minutes per patient, and is particularly helpful for slow typers.
It works by listening to conversations between medical professionals, or between doctor and patient, through a phone or a laptop. It transcribes the conversation, and then using AI, turns it into a summary which can then be entered into the hospital’s own system as notes for that particular patient.
New Zealand’s Breast Cancer Foundation says the use of artificial intelligence in screening will lead to cancers being discovered earlier.
Health Minister Simeon Brown confirmed in May that AI will serve as one of the two independent readers currently required in the mammogram assessment process from as early as next year.
The foundation’s chief executive Ah-Leen Rayner said AI should be seen as a second set of eyes and is proving its worth internationally.
In a letter outlining his expectations for Pharmac for over the coming year, New Zealand’s Associate Health Minister David Seymour has directed Pharmac to continue investing in data and digital infrastructure, including exploring the use of AI to support analytics, reporting and decisions across its core functions.
Seymour told RNZ’s Checkpoint that Pharmac had already made a positive start by using AI for tasks such as taking meeting minutes, summarising documents and categorising medical devices.
