Kia ora! Welcome to New Zealand’s weekly roundup of AI news, events, jobs and education.
McKinsey’s “State of AI 2025” says it all:
88% of companies use AI somewhere, yet two-thirds are stuck in pilots.
Only 23% are scaling AI in even one function.
Only 39% see measurable impact.
And that’s global. NZ lags even further…
Businesses are struggling because they don’t know where to start. Many are ticking the box without a plan:
A working group here.
A few Copilot licences there.
An intern testing prompts.
I’ve written a practical AI Adoption Framework for NZ businesses with 20–500 staff, based on real AI rollouts at companies I’ve worked at and others that are turning experiments into productivity gains. It’s built for clarity, not hype, helping leaders move from scattered tests to scalable capability.
If you want clarity and confidence on your AI direction before Christmas (and a head start on 2026) reply for a short, no-pitch conversation. We’ll share what’s actually working.
This week’s highlights:
Inside the Human Cost of AI
The Next Wave of AI Commerce
Health NZ Expands AI Note-Taking
The Five Pillars of AI Transformation
Microsoft’s Copilot Pricing Rollout Misfires
Fiona Hewitt: Leading Humans in the AI Era
Weekly Tip: Turn AI Into a Strategic Partner
Happy reading and listening ✌️
Did someone forward you this? Sign up!
HEADLINE STORY
⚠️ New Zealand's AI Adoption Snapshot
Global ranking: 21st place for AI adoption among developed nations
Individual usage: 38% of working-age adults currently use AI (Microsoft)
SME adoption: 68% of small and medium businesses have no plans to evaluate or invest in AI, citing lack of understanding and perceived absence of value
Large enterprise adoption: 80% of big firms are using AI tools
Economic potential: AI could add $76 billion to NZ's economy by 2038 (Microsoft/Accenture report)
Acceleration: AI Forum NZ reports adoption has accelerated significantly over the past six months, but we still lag behind other OECD countries like the US
Key barrier: Primary industry focus and competitive complacency — many businesses don't feel pressure to adopt because competitors aren't either
🏘️ Our Take: The $76 billion question is not whether AI will transform New Zealand’s economy. It is whether we will drive that transformation or watch it happen to us. A two-speed economy is emerging. Eighty percent of large firms are already in the game, while sixty-eight percent of our 500,000 SMEs remain on the sidelines. This is not a technology problem. The technology works. It is a learning crisis that will define New Zealand’s competitiveness for decades.
When most SMEs see no value in AI, that is a system failure in how we educate and support business owners. They are not avoiding AI out of caution; they simply do not know where to start. Business advisers, industry groups, and government programmes are not translating AI into language that resonates with builders, retailers, and service providers. The result is a widening productivity gap that risks locking local communities out of future economic growth.
The greater danger is complacency. Many businesses are not adopting because their competitors are not either, creating collective paralysis that holds everyone back. Meanwhile, international competitors are building AI-driven companies that will soon be impossible to match.
Being ranked twenty-first globally is not a disaster if we act as fast followers and learn from those ahead. But that requires active learning, not passive waiting. Our primary-industry base explains hesitation, not inaction. The irony is that the largest productivity gains are likely to come from the repetitive, manual work that dominates SME operations.
The shift from generative AI to agentic AI, where systems research, decide, and act autonomously, is already underway. Each month we delay widens the capability gap and erodes our future position.
New Zealand’s size can be an advantage. Smaller economies can pivot faster than large ones. But that will only happen if we invest in AI literacy that meets business owners where they are, not where technologists wish they were.
💼 Business & Industry
Microsoft's AI bundling strategy backfires in Australasia. Microsoft forced Copilot AI on subscribers, raising Family plan from $179 to $229. Legal pressure forced refunds and Classic plan disclosure.
2-min read.
Our take: The 28% price increase backfired spectacularly. When bundling AI becomes more costly than transparent, businesses lose trust faster than they gain revenue. This shows regulatory gaps in NZ consumer protection. Australia's ACCC acted; our Commerce Commission sent a warning letter.
Auckland wins 3,000-delegate AI medicine conference. Auckland secured MICCAI 2027, the world's leading medical AI conference, bringing 3,000 researchers from 60+ countries in September 2027. The event delivers $5.7M economic impact and 15,500 visitor nights.
2-min read.
Our take: Good sign to bring health-tech innovation to the country supporting our tech export market and increasing thought leadership locally.
⚖️ Government & Legal
Health NZ's AI scribe pilot goes national. The AI scribe trial we tracked is now live in 14 emergency departments, including Christchurch and Middlemore. Health NZ's new HealthX team is accelerating the rollout after positive clinician feedback.
12-min listen.
Our take: The productivity case is proven, Health NZ is moving fast from pilot to scale. The question shifts from "does it work" to "are we implementing it responsibly?".
Auckland academic warns lawyers on AI risks. Legal AI tools hallucinate 17-33% of the time, creating verification costs that erase efficiency gains, argues Dr Joshua Yuvaraj.
3-min read.
Our take: Yuvaraj raises valid concerns, but his critique leans on outdated models (17–33% hallucination rates) and overlooks the rapid evolution of legal AI. Modern systems using RAG and lawyer-in-the-loop workflows already show lower error rates than humans. The paper underplays where AI clearly adds value: from admin automation and knowledge management to large-scale regulatory and document analysis. The “verification paradox” is real, but review ≠ creation: validating AI output is typically 3–5x faster than drafting from scratch. The more interesting question isn’t if verification reduces value, but which legal tasks maintain strong ROI once verification is priced in.
🔍 Education & Society
Inside Big Tech's gig economy training your AI: one Kiwi's story on human cost. Johanna Knox trained generative AI for 18 months under exploitative conditions: wages slashed by 65%, team leads fired without warning, timeout penalties for platform glitches.
2-min read.
Our take: This is the invisible workforce behind every chatbot. Silicon Valley wants us focused on AI's existential risks while quietly building an empire on precarious labour.
📚️ AI for Business
Helping leaders and teams adapt, learn, and scale with AI.
1️⃣ Most AI pilots don’t fail. Leaders do. MIT said 95% flop, Wharton found 75% deliver ROI. The gap shows how alignment and execution separate hype from impact.
1-min read.
2️⃣ Google just rewrote how research gets done. Gemini Deep Research now connects to Gmail, Drive and Docs, turning company knowledge into a competitive edge through indexing the gateway and backbone of the internet.
1-min read.
3️⃣ AI transformation isn’t one project. It’s five. Literacy, workflow, capability, governance and strategy all move together. Miss one and adoption collapses.
4-min read.
4️⃣ The next phase of commerce is converging. Three forces: collapsing funnels, conversational commerce, and agentic commerce; are reshaping how products are discovered and bought.
5-min read.
🎙️ The AI Corner Podcast
This week’s guest is Fiona Hewitt, founder of Unboxable. Hear:
Why AI isn’t the headline: it’s the way we work that’s changing.
How control has become a liability for modern leaders.
What it means to be the last generation leading human-only teams.
Why NZ’s low AI confidence should fuel us, not hold us back.
🌍 Tech Updates From Global
The selected top headlines from each major AI tech company.
OpenAI
Secured a seven-year, $38 billion partnership with Amazon Web Services to access extensive NVIDIA GPU infrastructure for AI model training.
Reached over 1 million paying business customers with 7 million seats for ChatGPT for Work, becoming the fastest-growing enterprise AI platform in history.
CEO Sam Altman stated the company is on track to achieve an annualised revenue run rate exceeding $20 billion by year's end.
Launched its Sora AI video generator app on Android across multiple regions including the US, Canada, Japan, and Korea.
Unveiled Project Suncatcher, a space-based AI infrastructure using solar-powered satellite constellations equipped with TPUs to perform machine learning workloads in orbit, with first prototypes planned for 2027.
Released its first fully AI-generated advertisement produced with the Veo 3 video model, promoting Google Search's AI Mode across various media platforms.
Expanded Gemini's Deep Research capabilities for Gmail, Drive, and Chat, enabling users to generate AI reports referencing internal documents.
Anthropic
Reportedly in discussions with Google for a new funding round that could boost its valuation whilst offering additional cloud resources.
Microsoft
Made a substantial $15.2 billion investment in the UAE to bolster AI initiatives, positioning strategically in the growing regional AI market.
Received approval to ship over 60,000 NVIDIA chips to the UAE despite US export restrictions on advanced chips.
Established the MAI Superintelligence Team focusing on practical and high-impact AI applications.
Amazon
Rolled out generative AI shopping assistant Rufus across Europe after successful UK debut, enhancing customer shopping experiences.
Blocked Perplexity AI's Comet browser agent from accessing its marketplace, raising concerns about competitive practices in AI commerce.
Meta
Introduced the Vibes AI video feed to European users, expanding AI-enhanced social media capabilities.
Reportedly committed billions for up to 1 million TPUs as part of a partnership for low-latency, enterprise-scale inference.
Apple
Finalised plans to deploy a customised version of Google's Gemini model for its Siri overhaul, committing approximately $1 billion annually to license the technology ahead of anticipated spring release.
Perplexity
Secured a $400 million partnership with Snap to integrate its AI answer engine into Snapchat, expecting to generate revenue from the deal by 2026.
Signed a multi-year licensing agreement with Getty Images for AI-powered visual search to address criticism over data sourcing.
Miscellaneous
A coalition of Japanese creative organisations including Studio Ghibli formally requested OpenAI cease using their copyrighted works for training models, citing lack of consent.
💡 Tip of the Week: Turn AI Into Your Strategic Advisor
Most people still talk to AI like it’s Google. Smart people make it think like McKinsey.
LLMs already know how to reason, they just need structure to show it. By framing questions through trade-offs, constraints, and evaluation rubrics, you unlock the model’s ability to think strategically, not just summarise.
Why it matters
→ You stop getting generic lists and start getting executive-level reasoning.
→ You surface blind spots, second-order effects, and competing options.
→ You move from answers to decision frameworks.
This week’s exercise: Try one of these 10 prompt structures that forces AI to weigh options, score them, and challenge its own logic.
📅 AI Events in New Zealand
16 AI events happening across the country this week.
This week’s featured event:
Va’a Collective AI Wayfinding Workshop. This Saturday in South Auckland, join Pacific and Māori leaders to learn practical AI skills from them in a culturally safe space. Includes prompt book, kai, and follow-up session.
📅 Promote your event with us. Reply to let us know.
💼 AI Roles Around Aotearoa
Picklist of 🌶️ HOT 🌶️ new roles in AI this week.
Senior AI Engineer, Air New Zealand: Auckland
Business Analyst AI Insights, Deloitte: Wellington
AI and Automation Specialist, Ryman Healthcare: Christchurch
💼 Promote your job with us. Reply to let us know.
🤦 ️ AI Fail Of The Week
We all love AI, but it’s certainly far from perfect 🤔 …
👋 Mike & Erin