Why cuts first, AI later is the wrong move for New Zealand’s public service – Justin Flitter
The Government’s announcement of close to 9000 public service job cuts and artificial intelligence as the replacement is the most consequential AI decision New Zealand has made this year. It is also operationally backwards.
The sequence is wrong. Artificial intelligence (AI) does not add capability. It amplifies whatever capability is already there.
Cut 9000 people first and what AI amplifies is fewer people doing more work in workflows nobody redesigned, with less institutional knowledge, less context and a minister who could not say what the technology will cost when asked in Parliament.
Independent research published this month by Alethios confirms what we see daily across New Zealand businesses. Public-sector AI use is bifurcated, scattered and stuck in a pre-reasoning paradigm. Microsoft Copilot dominates formal use, while frontier models get used on personal accounts in shadow adoption. Senior leaders’ understanding of AI capability is anchored to 2023. A few power users get extraordinary value. Most do not.
That same pattern shows up across the firms we work with. Individual experimentation. Scattered tool adoption. Benefits to a few, occasionally.
The shift from ad hoc to embedded adoption requires an AI operating system. Not another tool licence. A system. The operating logic of how the organisation creates value with AI in the loop.
Build the context engine first. That is the domain knowledge, operational context, governance protocols and trust layer that let AI amplify capability instead of compounding risk. Then redesign the workflows. Then realise the efficiency. In that sequence, AI makes the public service more capable and more resilient, not just smaller.
As the 2026 Great New Zealand AI Roadshow departs Auckland this week for events in Nelson, Napier, New Plymouth, Tauranga, Hamilton, Wellington and Christchurch, the questions, the demand from businesses large and small, is direction, leadership and practical skills to confidently leverage AI, every day. The question has shifted from “Should we be using AI?” – they know they should – it’s now: “How can we deploy AI systems all our employees trust to delegate it their most tedious work?”
New Zealand has an unusually high AI appetite. Anthropic’s research puts Kiwi Claude usage at roughly four times the global per capita average, the fourth- highest in the world. The latent demand is real. The leadership layer has not caught up.
Government action is read as a signal by New Zealand business. If the public service models “cuts first, AI later”, businesses will copy that pattern, and we will watch the same shadow adoption, the same fragmentation and the same lost productivity show up across the private sector at scale.
Imagine the alternative. The Government uses this moment to model what good looks like. Pick three high-volume public sector workflows. Map how they should sense, think, decide and act with AI in the loop. Encode the trust layer and governance protocols. Train leaders to understand frontier capabilities, not just brand names. Publish the playbook. Make the operating logic visible.
That is not slower than the current plan. It is faster, because it would actually work. And it would give every Kiwi business a model worth following at exactly the moment most of them are wondering where to start.
The Minister is right that the public service is too fragmented and slow on digital. She is also right that AI presents an opportunity. What is wrong is the sequence. Cuts first assumes a capability that has not yet been built. AI does not fill that gap. It exposes it.
The harder, better play is to build the capability first, then realise the savings. That is the leadership New Zealand needs. That is the example our businesses are waiting to follow.
Operationally backwards is fixable. But only if we lead with the right question.
Not “which jobs can AI replace?”, but “how do we build the context engine and the operating logic that make AI a force multiplier for public service capability and resilience?”
That is the work. That is the leadership.
Catch up on the debates that dominated the week by signing up to our Opinion newsletter – a weekly round-up of our best commentary.