Job Displacement

The evidence base

AI and automation will not affect all jobs equally. Some roles face near-term displacement, others will be augmented, and some will remain largely unaffected. Understanding these differences is critical for targeted policy — and the NZ-specific data paints a sharper picture than many people realise.

Three major NZ-specific studies provide the foundation:

Reserve Bank of New Zealand (February 2026) — the most current assessment. Approximately 30% of NZ workers face high combined AI and robotics exposure. Less than 5% face low exposure — meaning almost no occupation is untouched. The critical finding: AI disproportionately affects professional, managerial, and administrative occupations in the $50,000–$150,000 income bracket. This reverses historical patterns where technology primarily displaced lower-skilled work.

NZ Productivity Commission (March 2020) — found 46% of NZ jobs face "high risk" of automation over 20 years. The Commission concluded that widespread disruption was unlikely within 10–15 years. That assessment was published months before GPT-3 and three years before ChatGPT. The timeline has compressed significantly.

Maxim Institute (February 2025) — estimated 40–60% of secretarial, administrative, and customer service roles are "exposed to and replaceable by AI today." Even a "modest" 10% elimination of all job roles within five years would cause "marked disruption in New Zealanders' lives."

What's already happening

The displacement is not hypothetical. An IDC survey of NZ organisations in 2025 found:

• 87% of NZ companies reported roles changing or disappearing due to AI in the past 12 months
• 53% reported complete role elimination — NZ had one of the highest rates globally, alongside the US and Argentina
• 34% have already slowed entry-level recruitment
• 88% expect further reductions in entry-level hiring within three years
• 76% report fewer on-the-job development opportunities for junior staff — the highest of any country surveyed

The dominant mechanism is not mass layoffs. It is the sinking lid: when people leave, they are not replaced. 40% of NZ companies report reduced need for new hires. Only 7% report AI directly replacing existing workers. The effect is the same — fewer jobs — but it happens quietly, one unfilled vacancy at a time.

NZ companies already affected

Datacom — NZ's largest technology services company. Cut 756 staff (12% of its workforce) in the year to March 2025, from 6,131 to 5,375. AI agents now write up to 70% of code in application modernisation projects, achieving 30–50% cost savings. Net profit rose from $34M to $37M despite the headcount reduction. CEO Greg Davidson called AI "the biggest revolution I've seen, maybe including the web."

NZ tech sector broadly — The TIN200 report showed total NZ tech sector revenue hitting $20 billion, while onshore staff fell 12.4% from 32,813 to 28,751 in a single year. Revenue growing while local headcount shrinks is the classic AI productivity–displacement pattern.

Health New Zealand — Slashed hundreds of IT roles — roughly a third of all digital positions — then had to hire Datacom consultants to cover the IT service desk after the restructure "gutted the department."

Xero — Cut up to 800 jobs in 2023, with a further 250 at risk in February 2026. Xero's core product is itself an automation tool that eliminates routine bookkeeping work, displacing accountants downstream.

Public Service — 3,044 FTE reduction from peak (December 2023), a 4.6% decline. While officially attributed to "government savings initiatives," multiple agencies are simultaneously deploying AI across claims processing, service delivery, and back-office functions.

Sector-specific exposure

New Zealand's service sector accounts for approximately 67% of GDP. Previous automation waves hit manufacturing and agriculture hardest — but LLM-based AI flips that pattern entirely.

Highest LLM exposure:

• Financial and insurance services — Credit scoring, underwriting, claims processing, compliance checking, and routine advisory work are all highly automatable. ANZ Bank in Australia has cut 4,500 staff; Macquarie warns the big four banks could replace 30% of staff with AI in 5–10 years. NZ's financial sector employs approximately 69,000 people.
• Professional, scientific, and technical services — Legal research, junior paralegal work, accounting, consulting analysis, and report writing. This is Wellington's largest industry at 14% of regional employment. Approximately 168,000 people nationally.
• Public administration — Nearly half the entire Public Service workforce is based in Wellington. Public servants comprise 23% of Wellington's overall workforce. Routine processing, policy drafting support, correspondence, and case management are all AI-exposed.
• Administrative and support services — Data entry, bookkeeping, scheduling, customer service. Approximately 66,000 people nationally.

Moderate LLM exposure:

• Retail — Customer service and back-office functions exposed, but in-store work less affected. Approximately 189,000 people.
• Education — Administrative functions exposed; teaching itself is augmented rather than replaced, but the nature of what's taught changes fundamentally.
• Healthcare — Diagnostic support, administrative tasks, and documentation are AI-augmented. Clinical care remains human-centred but the workforce mix shifts.

Lower LLM exposure (higher robotics exposure later):

• Agriculture, forestry, and fishing — Physical work is AI-resistant in the near term. Precision agriculture and autonomous systems are coming but on a longer timeline.
• Construction — Trades work requires physical judgment and dexterity that AI cannot replicate near-term. Design, compliance checking, and project management are more exposed.
• Manufacturing — Robotics exposure rather than LLM exposure. NZ manufacturing is already relatively small.

Regional variation

Wellington faces the highest concentration risk. Professional services and public administration together account for approximately 27.5% of Wellington's workforce — both are high AI-exposure sectors. If AI-driven restructuring hits government back-office functions and professional services simultaneously, Wellington's economy is disproportionately affected.

Auckland has the largest absolute numbers of workers in AI-exposed sectors but a more diversified economy, meaning lower concentration risk per capita.

Provincial New Zealand — paradoxically — may be more resilient to the first wave of AI disruption. Regions concentrated in agriculture, construction, and primary industries face lower LLM exposure. The 2015 NZIER report identified Canterbury, Waikato, Manawatu, and Otago as high automation-risk regions, but that analysis was based on robotics and routine task automation, not LLM-driven cognitive displacement. The RBNZ's 2026 paper distinguishes these two channels clearly.

This creates an unusual dynamic: the regions that felt "safe" in previous automation waves — urban centres with professional services economies — are now the most vulnerable. Rural and provincial NZ may find that its traditional economic base is, at least initially, a buffer.

Demographic impact

Gender: Women make up 61.4% of the public service — a heavily AI-exposed sector. Contact centre and administrative workforces also skew female. Women face disproportionate first-wave displacement, though the picture is complicated by lower female representation in the most senior (and most AI-resistant) management tiers.

Māori and Pasifika: The impact is uneven across different AI channels. Māori are overrepresented in labouring (16% vs 9% overall), machinery operation, and manufacturing — roles more exposed to future robotics than to current LLM automation. But Māori and Pasifika communities are also concentrated in customer service and contact centre roles that face near-term AI displacement. Only 2% of Māori workers are in financial services (vs 3% overall), providing some buffer against the highest LLM-exposed sector. The combined effect across all channels is still disproportionate risk, compounded by existing structural disadvantage in housing, education, and wealth.

Age: Entry-level workers face the steepest cliff. With 88% of NZ companies planning to reduce graduate hiring and 76% reporting fewer development opportunities for juniors, young people entering the workforce face a fundamentally different landscape than any generation before them. Globally, employment among 22–25 year olds in AI-exposed roles fell 16% from late 2022 to mid-2025. Older workers in senior positions are currently more insulated, but face retraining barriers if their roles are restructured.

Ethnicity and pay: At ANZ NZ, Māori or Pasifika employees are paid 21.9% less than European/Pākehā employees on average. Workers who are already paid less have less financial buffer to absorb periods of unemployment or retraining — making the displacement impact harsher even before considering the displacement probability.

International evidence

The displacement patterns emerging overseas are a preview of what NZ faces, compressed by 6–18 months:

Klarna (Sweden) — AI assistant doing the work of 700 full-time customer service agents within its first month. The company stopped hiring entirely and expects to shrink from 3,800 to 2,000 through attrition.

BT Group (UK) — Cutting up to 55,000 roles (42% of workforce) by end of decade, including approximately 10,000 replaced specifically by AI and automation.

Goldman Sachs — Estimates 6–7% of the US workforce (approximately 11 million workers) at risk with wide AI adoption.

UK graduate jobs — Job advertisements for occupations typical for graduates declined 33% since May 2022. King's College London found a 38% drop in AI-exposed role postings in 2024.

US software development — Developer job postings down approximately 70% from the 2022 peak. Entry-level postings down 60% between 2022 and 2024.

Freelance economy — Writing job posts on Upwork fell 30% after ChatGPT's release. More than half of businesses using freelancers in 2022 have stopped entirely. Business spending on labour marketplaces fell from 0.66% of total spend to 0.14%.

Commonwealth Bank (Australia) — Attempted to cut 45 customer service roles citing AI chatbots reducing call volume by 2,000 per week. Reversed the decision after union pressure when the AI "did not actually cut down the number of customers needing to speak to a human." A cautionary example of premature displacement.

Questions for contributors

1. What NZ-specific displacement data are we missing? The RBNZ 2026 analytical note likely contains occupation-level breakdowns — has anyone accessed the full paper?
2. Which NZ companies are actively reducing headcount due to AI but not announcing it publicly?
3. How are NZ unions responding to AI-driven restructuring? What protections are they negotiating?