December 2025
Our natural advantages: Integrating Te Ao Māori in Aotearoa AI
Building the AI world around people is essential, and when we apply diverse cultural lenses to AI challenges, we reap the rewards of a more connected approach to our community – which in turn allows us to move from commercially focused ESG models to the concept of “intergenerational AI”.
In Aotearoa New Zealand we see positive practical implications when the principles of data sovereignty and community engagement are applied through the integration of Te Ao Māori worldviews.
Platforms like Pera Barrett’s knowledge and learning platform “Kaha Create” are pioneering Māori data sovereignty, embedding genealogy tracking and community care into their processes, and hosting data locally to ensure regional sovereignty. Initiatives like this demonstrate how cultural perspectives can shape ethical AI development and governance for the benefit of us all.
Quantum Computing: Strategic Threats and Opportunities
2025 has been declared the International Year of Quantum Science by UNESCO, recognising quantum computing as both a major opportunity and a looming threat.
The economic disruption potential is significant
We must also be cognisant of the challenges quantum computing will bring. A year ago we thought the first serious impacts from Quantum would be 5-10 years away, but AI powered dev and test cycles alongside huge investments have reduced this expectation down to to 2-3 years.
While standard computers run solution scenarios in series, a quantum computer can run every scenario simultaneously, speeding up computing exponentially. For example:
- A classical computer trying to break a 2048-bit RSA encryption key (the standard for banking and secure websites) would take roughly 300 trillion years vs a quantum computer which would do it in hours.
- Simulating a single caffeine molecule on a classical supercomputer would require more bits than there are atoms in the observable universe (~10⁸⁰). A quantum computer could simulate it with just a few hundred qubits.
In practical applications, a delivery company like UPS or FedEx planning routes for 100 trucks faces around 10¹⁵⁷ possible paths — an astronomically large number. Quantum computing could analyse all these routes in parallel, identifying the optimal one up to 100 million times faster than classical optimization models.
The impact on Gen AI will be astounding. While training a large AI model (like GPT-4) can take weeks on thousands of GPUs, quantum machine learning (QML) algorithms could cut training time by 90%, thanks to parallel state exploration in high-dimensional spaces.
While the scientific breakthroughs hold great promise, the threat of misuse is game changing. The potential for quantum computers to break existing encryption within 3–5 years, and perhaps with the first instances inside 2 years, leaves current encryption techniques and decades of archived hacked data vulnerable.
What are we doing about it? Aotearoa New Zealand lacks a post-quantum cryptography strategy, with only banks currently preparing for this shift, making it one of the Forum’s focuses for 2026.
Our aim is to create knowledge and awareness of the impact of quantum within our own Working Groups and subcommittees, and more broadly to promote coordinated standards and policies, and extend the integration of quantum topics into skills and talent programmes.
Trust and safety enable AI innovation to go faster
In an industry and social licence, it’s worth recognising that past profound technological shifts have been enabled by trust and safety allowing innovation to move faster. And the same thing is possible with AI.
Electricity is inherently dangerous, and in 1889 people talked about the “unrestrained demon of electricity”. Understandably consumers would not touch light switches for fear of electrocution – even US President Benjamin Harrison hired an electrician to turn the White House lights on and off for him and his wife. Today we have a whole range of laws, standards and codes that enable us to trust that electricity is safe to use in our homes and businesses.
The first cars had a top speed of 16 km/h until Bertha Benz’s invention of modern braking enabled cars to go faster – because they could now stop safely. Similarly, Mr Elisha Otis’ invention of a mechanical safety device in lifts enabled the building of skyscrapers by transforming lifts from dangerous freight platforms into trusted passenger systems.
Given trust and safety guardrails have a proven history of enabling innovation to go faster, isn’t it time to start thinking about AI in the same way? Trust can be fostered through clear standards, industry collaboration and proportionate guardrails that give the public confidence while balancing the need for innovation and investment.
Green AI & Sustainability: Rethinking Research Models
Sustainability in AI is a twofold issue: Firstly, using AI for environmental benefit and secondly developing energy-efficient methods for delivering AI.
As a response to both of these concerns we are seeing a current trend towards distributed computing (cloud to edge), with local processing proving better for efficiency and privacy; and a move away from large centralized systems towards narrow focused, specialized solutions capable of servicing the tasks at hand.
This trend towards distributed computing—from cloud to edge—promises greater efficiency and privacy and will provide part of the answer to AI’s environment sustainability going forward. However, current major costs are proving a barrier for AI research, which in turn is driving a fundamental change in approach.
The high barriers threaten to kill early-stage research, with New Zealand getting left behind and any benefits flowing overseas. For example, the cost of training and GPU costs for large models like LLama 3 ($20-50M) now exceed traditional supercomputer advantages. This is encouraging a move from individual projects to collaborative team efforts and is reflected in the approach NZIAT has taken to $70m AI platform funding
Digital Twins & Data Integration: People-First Innovation
Smart cities are people-first environments enabled by technology.
Digital Twins are not new, but AI is driving the evolution from isolated digital twins to interconnected systems, enabling real-time data interrogation and decision-making, including the 2025 Aotearoa AI Award winners ALMA at PHFScience [insert link].
AI enables interconnected digital twins vs isolated replicas with real-time data interrogation and decision-making; and reduced costs vs traditional digital twin approaches
Success stories include flood prediction in Bangkok reduced from 30 hours to 30 seconds processing time; a 30% traffic reduction for Toronto Air Traffic control; and water quality monitoring in Auckland providing real time safe to swim advice. The Moata platform [insert link], born in New Zealand, exemplifies how rapid iteration and minimal bureaucracy can lead to global deployment.
The holistic views that interoperability of digital twins promise enhanced future capabilities for disaster planning, responses and management; better planning decisions and reductions in costs for infrastructure projects; and better response to epidemics and pandemics – in both animals and humans. All of which would have profound economic impacts.
Thief or genius? AI’s Contentious role in the Creative Sector
The relationship between different creative disciplines and AI remains contentious, with problematic binary positions of “AI replaces creativity” verses “AI enhances business operations”.
The global legal landscape has seen a large number of IP and cultural appropriation (or theft) lawsuits with mixed outcomes. To date plaintiffs have been generally unsuccessful with AI-generated works increasingly being recognised as “fundamentally new”. However, defendants are still paying settlements as the inter-relationships between the IP used as training data, the outputs, and their use (e.g. for a competitive product) are complex.
While many issues remain, especially with regards to cultural appropriation, the impact of these legal cases has been profound at least for larger global players with many now paying for IP as a competitive advantage. Revenue-sharing models with publishers, including OpenAI with NewsCorp and Microsoft with Taylor Francis, are creating high quality data moats that drive users back to the publishers, creating a sustainable mutually beneficial ecosystem.
From a creator / artist perspective, AI is also providing both support for creators and creating new itself:
- There are clear opportunities in administrative support for our 300,000 creative sector participants, most of whom are sole traders handling marketing, grants, publicity and administration for which AI is well suited to act as an assistant.
- AI is also well recognised as an assistant in the creative process itself. Speeding up tasks and reducing rework by creating mock ups of visuals, reduces both time and cost of production for individual creators.
- AI can also be seen as a creative medium in itself, with examples including Refik Anadol creating works that turning museum data into visual art installations; and Aida robot’s $1.1M artwork sale (albeit with a team of 6 engineers/artists behind her).
AI Learning & Research: Accelerating Discovery
AI is dramatically accelerating research and development cycles, from software engineering to scientific discovery.
The collaborative potential of AI-human co-improvement calls for, advocating for a partnership approach rather than letting AI develop independently.
The research cycle—definition, ideation, experimentation, learning—is being compressed, and AI tutoring offers new opportunities, though safety considerations remain paramount.
NZ AI Adoption: Trust, Productivity, and Governance
Looking at findings from NZ AI adoption surveys and reports, we can see that, while AI use is becoming widespread, it is surprisingly difficult to pin down exactly how much AI is being used. A read across of 28 reports provides wildly varying adoption rates (40-80%) depending on survey methodology, with high levels of adoption amongst corporate workers and lower levels across the broader population. Information workers have the highest levels of adoption (higher than global average). When taken at face value, this can make the national picture look confusing or even contradictory.
When these differences are accounted for, a more consistent story emerges. Many New Zealanders now encounter AI in their daily lives, and a growing number are using it at work in practical ways, particularly for routine tasks and information processing. More intensive or deeply integrated use is less common, but increasing. Importantly, across almost all studies, people and organisations consistently report positive productivity effects, including time savings and efficiency gains, even where overall use is still relatively light or informal.
At the same time, trust has not kept pace with uptake. Analysis shows that concerns about privacy, security, bias, job impacts, and transparency are persistent, including among people who already use AI tools. In this sense, New Zealand can be described as a high‑use but low‑trust environment for AI. The findings suggest that the next phase of AI adoption will be shaped less by access to technology, and more by how well organisations, government, and society address trust, skills, and governance alongside continued use.
Workforce Transformation: AI as Collaborator
In the last 2 years AI has shifted from assistant to genuine collaborator, with key opportunities in augmented decision making, productivity gains through manual task reduction and T-shaping of roles, lowering barriers to adjacent skills.
With this rapid pace of change, change fatigue and lagging infrastructure integration needs are emerging as key challenges. Meanwhile, as traditional technical boundaries are collapsing, new roles are emerging from AI product owners and architects to model ops specialists. This plays to our natural advantages of a workforce that is already skilled at adaptability and wearing multiple hats.
Conclusion
The AI landscape is rapidly evolving, culturally nuanced, and has both opportunity and risk. Success will depend on integrating diverse perspectives, preparing for new forms of disruptors like quantum, rethinking research models, and fostering trust and governance to support responsible and inclusive innovation.