AI in Academic Research in Aotearoa New Zealand: A Living Whitepaper
Last updated: 2 April 2026
Introduction
AI adoption in New Zealand’s academic research sector has continued to deepen since the previous update on 9 March 2026, but the most important shift is not simply that more researchers are using more tools. The sector is increasingly being shaped by platform-building, institutional governance, secure tool provisioning, and translational research programmes that connect universities, Public Research Organisations, funders, and industry. The clearest new signal is that the national AI Research Platform process has moved from broad ambition to a live contest between five distinct consortia, with final decisions expected in late May 2026. (mbie.govt.nz)
Executive Summary
- The national AI research race has entered a decisive stage. MBIE says more than 100 concepts were submitted in Phase 1, with five concepts selected to progress; each received NZ$250,000 to develop a full proposal, and the final platform decision is due in late May 2026. The shortlisted ideas now span agentic AI, creative AI, autonomous systems, bioeconomy AI, and outdoor/physical AI. (mbie.govt.nz)
- The biggest post-9 March development is the emergence of “Outdoor AI” as a visible contender. On 31 March 2026, the Universities of Waikato and Canterbury publicly outlined a joint proposal focused on AI systems that can operate in complex real-world environments across agriculture, horticulture, aquaculture, and related sectors. (waikato.ac.nz)
- Governance is tightening and getting more operational. Otago’s university-wide AI Governance Policy took effect on 10 March 2026; UC published new researcher-facing AI tool and responsible research guidance in February–March 2026; and funder guidance now more explicitly addresses how generative AI may and may not be used in proposal preparation and assessment. (otago.ac.nz)
- Health remains the strongest adoption cluster, but it is no longer the only one that matters. HRC’s 2025 AI in Healthcare initiative is now confirmed in HRC’s 2025 annual reporting as 10 funded studies worth NZ$4.6 million, while major AI-led activity is also visible in climate modelling, aquaculture, digital twins, autonomy, and bioeconomy research. (hrc.govt.nz)
- University-level AI access is becoming more structured. The University of Auckland provides secure access to Microsoft 365 Copilot Chat, Google Gemini, and NotebookLM; UC says it has procured secure versions of ChatGPT and Microsoft Copilot for research use and explicitly warns against putting in-confidence or sensitive data into public tools. (auckland.ac.nz)
- The sector still shows strong activity but incomplete measurement. There is now abundant evidence of funded projects, policies, and infrastructure, but there is still no clean nationwide benchmark of AI adoption across all research disciplines and institutions. That remains a structural gap. This is an inference from the current evidence base rather than a directly reported metric. (mbie.govt.nz)
What Changed Since the 9 March 2026 Version
1) Otago moved from draft-era governance to active policy
A meaningful change since the prior version is that the University of Otago AI Governance Policy is no longer just an announced framework; it took effect on 10 March 2026 and applies to AI development, deployment, procurement, and use across the institution, including research activity. The policy explicitly frames AI governance as something that must remain flexible because risks, opportunities, and best practice are evolving rapidly. (otago.ac.nz)
2) UC made research-facing AI enablement more explicit
The University of Canterbury published a new Responsible research page on 3 March 2026 that places AI use alongside ethics, Māori consultation, intellectual property, and responsible conduct. It also published a dedicated Research and AI Tools page on 17 February 2026, describing endorsed tools, confidentiality constraints, and the availability of secure institutional options for ChatGPT and Microsoft Copilot. (canterbury.ac.nz)
3) The AI Research Platform story broadened beyond one leading proposal
The previous version highlighted the University of Auckland-led Aotearoa Agentic AI Platform. Since then, the platform landscape has become more concrete and more plural: MBIE’s December 2025 shortlist confirms five Phase 2 concepts, and the Waikato–Canterbury Outdoor AI proposal has now publicly entered the conversation as a nationally significant alternative vision. (mbie.govt.nz)
4) Marsden guidance has become more nuanced
The 2026 Marsden Fund EOI guidelines continue to warn applicants to use caution with generative AI and require them to take full responsibility for proposal content, originality, and source validity. At the same time, draft 2026 guidance for council and panel members allows limited use of AI tools such as Research Rabbit for referee finding, provided confidential proposal information is not entered and outputs are checked for accuracy and bias. That is a notable maturation from blanket caution toward controlled, bounded use. (royalsociety.org.nz)
Current State of AI Adoption
High-level snapshot
As of 2 April 2026, AI adoption in New Zealand academic research is best described as institutionalising, mission-led, and still uneven. It is strongest in settings where there are rich datasets, high-value prediction or simulation tasks, translational partners, and access to specialist infrastructure. Health remains the densest cluster, but the visible frontier now also includes climate science, aquaculture, autonomous systems, digital twins, and bioeconomy research. (hrc.govt.nz)
What has matured most is the scaffolding around AI, not just the tools themselves:
- national platform competition and strategic investment design, (mbie.govt.nz)
- integrated eResearch infrastructure through the REANNZ–NeSI platform, (mbie.govt.nz)
- explicit research integrity and transparency guidance from Royal Society Te Apārangi and Marsden, (royalsociety.org.nz)
- formal ethics oversight for health research using AI or algorithmic tools, (ethics.health.govt.nz)
- and university-level policies that increasingly link AI use to privacy, sovereignty, accountability, and secure access. (aut.ac.nz)
The overall pattern suggests that New Zealand academic research is moving from isolated AI projects toward a more coordinated research system in which AI is treated as both a scientific capability and a governed institutional resource. That is an inference, but it is strongly supported by the convergence of funding, infrastructure, ethics processes, and university controls. (mbie.govt.nz)
Latest News and Strategic Developments
1) The national AI Research Platform is now the central strategic story
MBIE’s Artificial Intelligence Research Platform has become the most important single development in the sector. Phase 1 has now concluded, with five concepts funded to progress to Phase 2. MBIE states that each concept received NZ$250,000, with Phase 1 contracts finishing on 31 March 2026, and that the platform is intended to receive up to NZ$70 million over seven years. (mbie.govt.nz)
The shortlisted concepts reveal how broad New Zealand’s AI research ambitions have become:
- University of Auckland: Aotearoa Agentic AI Platform, focused on next-generation autonomous agents and simulation-based testing. (mbie.govt.nz)
- Wētā FX: Aotearoa Creative Artificial Intelligence Research Institute, focused on computer vision, generative models, digital twins, and AI rights management. (mbie.govt.nz)
- Earth Sciences New Zealand + Victoria University of Wellington: Aotearoa Institute for Autonomous Intelligence, focused on advanced AI and autonomy in aerospace and marine settings with spillovers to agriculture, logistics, energy, and environmental monitoring. (mbie.govt.nz)
- Bioeconomy Science Institute: BioAI Platform, focused on agriculture, aquaculture, forestry, pest threats, and climate resilience. (mbie.govt.nz)
- Waikato + Canterbury: a merged Physical AI / Outdoor AI concept for intelligent systems that sense, learn, and act in real-world outdoor environments. (mbie.govt.nz)
2) The latest live contender is Outdoor AI
The most significant new public announcement since the previous version is the 31 March 2026 Waikato–Canterbury release on Outdoor AI. The proposal focuses on physical and digital AI systems designed for complex real-world conditions, with initial applications in agriculture, horticulture, aquaculture, and other primary sectors. The announcement also says results are due in late May 2026, making this a near-term decision point for the sector. (waikato.ac.nz)
This matters because it broadens the national debate from “AI for productivity” in the abstract to a sharper question: what kind of AI should New Zealand specialise in? The shortlisted concepts imply several competing answers, from agentic AI and creative AI to outdoor autonomy and bioeconomy AI. That diversification is one of the clearest signs of sectoral maturity. (mbie.govt.nz)
3) Digital research infrastructure is being aligned to AI use
MBIE’s REANNZ–NeSI integration, announced on 5 August 2025, remains foundational. The unified platform is intended to provide shared access to high-performance computing, AI capability development, specialist data-network access, and researcher training, backed by NZ$13.93 million annually through the Strategic Science Investment Fund. (mbie.govt.nz)
At the institutional level, universities are now making AI access more operational. The University of Auckland says it provides secure access to Copilot Chat, Gemini, and NotebookLM, and explicitly positions NotebookLM as an AI-powered research assistant for working with users’ own documents and sources. The University of Canterbury similarly says it has procured secure versions of ChatGPT and Microsoft Copilot and warns researchers not to put in-confidence, sensitive, or special-classified data into public tools. (auckland.ac.nz)
Research, Governance, and Institutional Readiness
National guidance is now substantive, not symbolic
Royal Society Te Apārangi’s guidance remains the clearest sector-wide reference point for GenAI in research. It says researchers are accountable for AI-assisted outputs, that GenAI systems are not authors or co-authors, and that substantial use should be transparently indicated, including the tool used and how it affected the research process. The guidance also foregrounds Māori and Pacific data sovereignty, privacy, confidentiality, intellectual property, hallucinations, and bias. (royalsociety.org.nz)
The Marsden Fund has carried this logic into funding practice. The 2026 EOI guidelines caution applicants about authorship, copyright, and factual accuracy issues, and require them to confirm that they take full responsibility for proposal content and cited sources. Draft 2026 panel guidance introduces a narrow exception: panellists may use AI search tools for referee-finding, but must not enter confidential proposal content and must review outputs for appropriateness and bias. (royalsociety.org.nz)
Ethics oversight is now procedural
The Health and Disability Ethics Committees require a supplementary form whenever AI or algorithmic tools are used or developed in a health research project. Applications that include AI components but omit the form are sent back for completion. This is an important sign that AI is now embedded in ethics workflow rather than handled as an unusual edge case. (ethics.health.govt.nz)
University governance continues to harden
Recent university signals show a clear move from experimentation toward governed adoption:
- AUT’s January 2026 research guidance is explicitly framed as a living document and emphasises researcher responsibility, ethics compliance, transparency, and Māori and Pacific data sovereignty. (aut.ac.nz)
- Otago’s AI Governance Policy took effect on 10 March 2026 and applies across AI development, deployment, procurement, and use, including research. (otago.ac.nz)
- Massey’s staff guidance says GenAI can support research and accelerate discovery, while stressing privacy, security, equity, Māori data sovereignty, and research conduct risks. (massey.ac.nz)
- UC’s research-facing AI guidance explicitly links AI use to confidentiality, ethics approval, commercialisation risk, attribution, and secure institutional tooling. (canterbury.ac.nz)
- The University of Auckland has gone furthest in standardising secure access to named tools for staff and students. (auckland.ac.nz)
Research and Case Studies
Case Study 1: The AI Research Platform competition is becoming a de facto map of national priorities
The five shortlisted AI Research Platform concepts together offer the best current map of where New Zealand thinks AI-enabled research advantage may lie: agentic systems, creative AI, edge autonomy, bioeconomy AI, and outdoor/physical AI. Because these are not isolated grants but competing bids for a national platform, they show where capability, commercialisation ambition, and sector alliances are coalescing. (mbie.govt.nz)
For academic research specifically, this is important because the shortlist joins universities, Public Research Organisations, iwi-linked interests, and firms in larger consortia rather than project-by-project activity. That indicates a move toward institutional concentration and specialisation. This is an inference, but it is well supported by the composition of the shortlisted proposals. (mbie.govt.nz)
Case Study 2: Health research remains the leading edge
HRC’s 2025 annual reporting confirms that its AI in Healthcare initiative funded 10 studies worth NZ$4.6 million, from a NZ$5 million pool, and that the call attracted many applicants who were new to HRC. That is one of the strongest available indicators that AI is pulling new actors into parts of the research system. (hrc.govt.nz)
The funded portfolio shows breadth as well as depth:
- AI-assisted stroke care aims to improve rapid scan interpretation and could enable an additional 850 New Zealanders per year to access critical treatment, with projected health-system and societal savings if implementation succeeds. (hrc.govt.nz)
- AI-enhanced digital pathology at Otago is targeting better prognosis and treatment decisions for gastrointestinal cancers. (hrc.govt.nz)
- Explainable AI for dementia risk at the University of Auckland is using the Stats NZ Integrated Data Infrastructure to identify modifiable risks, with transparency and inequity reduction built into the work. (hrc.govt.nz)
- Aotearoa-specific AI evaluation framework work is trying to build local methods for assessing AI in health rather than importing overseas assumptions uncritically. (hrc.govt.nz)
- Ethics guidelines for AI in youth mental healthcare are being co-designed with young people, clinicians, Māori communities, and government agencies. (hrc.govt.nz)
Together, these projects show that health AI in New Zealand academia is moving beyond model development into implementation, evaluation, workflow redesign, and culturally grounded governance. (hrc.govt.nz)
Case Study 3: AI for healthy ageing has become a nationally visible translational cluster
MBIE’s New Zealand–Singapore Leveraging AI for Healthy Ageing 2025 initiative remains one of the country’s most concentrated AI research pushes. The programme funds three projects, each with NZ$4 million for the New Zealand partner over three years. (mbie.govt.nz)
The projects remain strong examples of translational academic research:
- University of Auckland: an AI-driven dementia risk score combining multimodal health data. (mbie.govt.nz)
- University of Otago: AI-InterRAI, aimed at partially automating assessments and improving personalised care planning. (mbie.govt.nz)
- Victoria University of Wellington: AIMCura, a mobile-first AI-augmented system for remote cognitive monitoring and support. (mbie.govt.nz)
Case Study 4: Climate, digital twins, and physical systems show AI spreading beyond health
At Earth Sciences New Zealand / NIWA, researchers are developing a physics-informed generative AI method for high-resolution climate projections that the organisation says is more than 1,000 times faster than current physics-based methods. The work has already been linked to a published Geophysical Research Letters output and is being positioned as a way to make large high-resolution ensembles feasible for New Zealand climate-risk analysis. (niwa.co.nz)
In bioengineering, MBIE announced on 18 September 2025 up to NZ$4.5 million over four years for a partnership between the Auckland Bioengineering Institute and the University of Texas at Austin to build AI-enabled digital twins. MBIE says the project could support personalised healthcare, drug development, agricultural innovation, hospital savings, and licensing revenue. Separately, MBIE’s Horizon Europe top-up funding page confirms Auckland’s participation in the VITAL project on virtual twins for personalised clinical care. (mbie.govt.nz)
At the University of Canterbury, AI-enabled aquaculture research is using autonomous underwater vehicles, imaging, and 3D reconstruction, with UC saying the immediate benefits to the mussel industry alone could be worth up to NZ$80 million per year. (canterbury.ac.nz)
These examples show a second major adoption pattern in New Zealand academia: AI is increasingly being used where simulation, sensing, robotics, and environmental complexity overwhelm conventional workflows. (niwa.co.nz)
Emerging Trends
1) The sector is shifting from “tool use” to “research system design”
The strongest evidence is no longer about whether researchers have access to AI chat tools. It is about whether institutions can support secure, reproducible, ethics-aware, domain-specific AI workflows. The rise of platform proposals, evaluation frameworks, digital twins, climate emulation, and secure university tooling all point in that direction. (mbie.govt.nz)
2) New Zealand-specific values are shaping adoption
Across Royal Society guidance, university documents, and platform proposals, AI adoption is being framed through Te Tiriti obligations, Māori data sovereignty, Pacific values, community benefit, and local relevance rather than through generic international AI language alone. (royalsociety.org.nz)
3) Commercialisation pressure is becoming more visible
The national platform competition is explicit about building durable competitive advantage, knowledge-intensive firms, and research-to-market pipelines. That means academic AI research in New Zealand is increasingly being judged not only by scientific quality, but by its ability to create firms, productivity gains, and exportable capabilities. (mbie.govt.nz)
4) Health leads, but the frontier is now multi-sector
Health still has the deepest combination of funding, ethics processes, and translational case studies. But the Phase 2 platform shortlist, NIWA’s climate work, UC’s aquaculture programme, and the digital twin investments show that AI adoption is now clearly multi-sector. (hrc.govt.nz)
5) Measurement still lags activity
There is now strong evidence of activity, but still weak system-wide measurement of researcher adoption by discipline, maturity, or workflow type. New Zealand can point to grants, policies, and infrastructure; it still struggles to point to a single authoritative baseline adoption dataset across the academic research system. (hrc.govt.nz)
Overall Assessment
As of 2 April 2026, AI in New Zealand academic research is best understood as institutionalising rather than saturating. The sector is not uniformly AI-native, and broad adoption measurement remains incomplete. But the direction is now unmistakable: AI has become a serious object of national research investment design, ethics oversight, university governance, secure tooling, and translational programme building. (mbie.govt.nz)
The most important change since the 9 March 2026 version is that the story is no longer mainly about one promising platform or a handful of funded projects. It is now about a contest to define New Zealand’s AI research specialisation, with visible pathways in health, agentic AI, outdoor autonomy, creative AI, bioeconomy AI, climate modelling, and digital twins. (mbie.govt.nz)
The strongest synthesis is this: New Zealand academic research is not treating AI merely as a writing aid or generic productivity layer. It is increasingly treating AI as a governed national capability—one that must align with research integrity, Te Tiriti-informed responsibilities, data sovereignty, and pathways to real-world impact. If the AI Research Platform delivers as intended after the late May 2026 decision point, the next phase is likely to be more concentrated, more collaborative, and more internationally visible than the sector’s still-fragmented project base today. (royalsociety.org.nz)