Year 5–10: Structural Transformation (2031–2035)

Context

By 2031, the question is no longer whether AI is transforming the economy — it will be — but whether NZ's institutions are capable of governing that transformation in the interest of all New Zealanders. This phase tests that capability. The decisions made in Years 1–5 will have either built the foundation for structural change or not. If they have, Year 5–10 is where the deeper transformations become possible. If the foundation is inadequate, this phase will reveal that too — and the adaptive policy cycle must respond.

Measuring Against the Baseline

The National AI Impact Assessment commissioned in Year 1 is the baseline against which 2031 outcomes should be measured. By this point, NZ should be able to answer concretely: which sectors saw the displacement levels the model projected? Where did outcomes differ from projections, and why? Which interventions produced measurable improvements in employment outcomes, income security, or skills attainment — and which did not?

This is not retrospective accountability for its own sake. It is the closing of an evidence loop that feeds directly into the next phase of the AI Commission's five-yearly pathway recommendations. Governments that governed well in Years 1–5 will be able to demonstrate measurable outcomes. Those that did not will face the data. This accountability structure is one of the strongest arguments for investing seriously in the Year 1 assessment.

If the labour market data by 2031 shows persistent structural displacement that wage subsidies and retraining alone cannot address, the case for trialling Universal Basic Income becomes substantially stronger. NZ is well-positioned to run a rigorous UBI pilot — small enough to be manageable, with strong statistical capacity and a tradition of evidence-based social policy. A pilot covering 5,000–10,000 participants across a representative range of demographics and regions, running over three years, would generate world-class evidence on UBI's effects on employment, wellbeing, and social cohesion.

The pilot should be designed before it is needed — in Years 3–5 — so that if the Year 5 labour market data triggers the threshold for implementation, it can begin without a further two-year design phase. Pre-designing does not commit NZ to running the pilot; it preserves the option.

Mature Sovereign AI Infrastructure

By 2031, if the sovereign compute investment was made in Year 3, NZ should have operational AI infrastructure supporting a significant share of sensitive government workloads. The maturation question is not primarily technical but institutional: does NZ have the procurement capability, the technical workforce, and the governance frameworks to operate this infrastructure at the standard that public trust requires?

The technical workforce question is particularly important. Sovereign AI capability requires a pipeline of New Zealanders with relevant skills — not just data scientists but also AI safety specialists, policy analysts with technical literacy, and public servants who can effectively commission and oversee AI systems. The education reforms of Years 1–5 should have begun building this pipeline, but Year 5–10 is where the pipeline's adequacy becomes testable against actual institutional demand.

NZ as an AI Governance Exemplar

A country of five million people cannot compete with the US or China on AI capability. It can compete on AI governance — and governance is increasingly where the global stakes are highest. By 2035, NZ's goal should be to be recognised internationally as a model for how a small, open, democratic economy manages AI transition equitably and sustainably.

This is not a vanity objective. Being a recognised governance exemplar gives NZ disproportionate influence in international standard-setting, attracts researchers and companies seeking stable regulatory environments, and builds the trust internationally that makes NZ's voice heard in bodies like the OECD, the UN, and the Global Partnership on AI. It also creates domestic accountability: a country that positions itself as a governance leader faces higher scrutiny if its domestic outcomes are poor.

The test of governance exemplar status is outcomes, not process. What are NZ's Gini coefficient trends? What happened to employment rates in high-exposure sectors? How did the education reforms affect equity gaps? What did the Citizens' Assembly recommend, and which recommendations were implemented? These are the questions that international observers will ask, and they are the right questions.

The Adaptive Policy Cycle

The roadmap ends at 2035 but the challenge does not. AI capability development shows no signs of plateau, and the second-order effects of AI on social organisation, democratic participation, and geopolitical power are only beginning to be understood. The most important institutional outcome of this decade is not any specific policy but the establishment of an adaptive policy cycle — the permanent institutional capacity to hypothesise, pilot, measure, and adapt in response to a rapidly changing technological environment.