AI & The Future of Work
The dominant media narrative — that AI will trigger mass layoffs — is largely wrong, at least for now. What is actually happening is more insidious: AI is removing the bottom rungs of the career ladder, silently eliminating entry-level and junior roles that have historically served as training grounds. Junior accounting clerks. Data entry operators. Tier-1 call centre agents. These roles are not being axed in dramatic announcements — they are simply not being refilled when people leave.
In New Zealand, the IDC/Deel Report 2025 found that 34% of NZ companies have already slowed entry-level hiring due to AI, with 88% expecting to do so within three years. Youth unemployment (ages 15–24) has climbed to 15.2% — more than three times the 4.8% national average — with the divergence accelerating sharply from late 2023, coinciding with mainstream generative AI adoption across the knowledge economy.
A critical aggravating factor is work arrangement. The Warwick/Oxford 2026 study found that remote and hybrid roles carry three times the AI displacement risk of fully on-site roles — because they concentrate information-processing tasks that AI can replicate. New Zealand's post-COVID acceleration of hybrid work in the public service and financial sector has inadvertently moved a large cohort into the highest-risk AI exposure bracket.
“AI is not replacing workers. It is replacing the jobs that teach people how to become workers — and New Zealand's graduates are the first to feel the draft.”
— Synthesised from Oxford Future of Work Institute findings & IDC/Deel NZ Report, 2025–2026One of the most counter-intuitive findings of recent research: the shift to remote and hybrid work has created a structural vulnerability to AI displacement. Remote work concentrates employees in document handling, email, scheduling, data analysis, and reporting — precisely the task categories most susceptible to AI automation. The Warwick/Oxford 2026 study quantifies this with striking clarity, and the implications for post-COVID New Zealand are significant.
Over 80% of NZ organisations are now using AI in some operational capacity, yet mass layoffs remain rare. The IDC/Deel 2025 survey found that while 53% of NZ organisations report roles have disappeared due to AI, only 7% report mass redundancy events. The mechanism is attrition and vacancy elimination: when someone leaves, the position is simply not refilled. The work is absorbed by AI tools and remaining staff — invisible to headline statistics, but cumulative and structural.
Financial services, legal, and insurance have been most aggressive in NZ. The telecommunications sector has seen the sharpest contraction, with AI-assisted triage handling an estimated 40–60% of inbound customer contacts at major carriers without human intervention. The largest displacement effects are occurring in large enterprise and government — organisations with the IT infrastructure and capital to deploy AI at scale.
This is creating a two-speed labour market: large employers doing more with fewer people, while SMEs struggle to compete for the remaining talent. The gap is structural, not cyclical, and will widen as AI tools become cheaper and more capable.
A mid-sized Wellington contact centre serving a national insurance client — approximately 180 FTE — began a phased AI deployment in Q2 2023. Phase 1 deployed a conversational AI triage system resolving 44% of inbound contacts by Q4 2023. Phase 2 introduced AI-assisted agent tools, cutting average handle time on complex queries by 38%.
No staff were made redundant. When 40 agents left through natural attrition over 18 months, only 16 were replaced — a net reduction of 24 FTE (13%). The remaining team handled 12% higher volume. The client reported a 22% reduction in service costs against a 15% improvement in customer satisfaction.
The centre's HR manager described it as "entirely painless from a management perspective" — no redundancy payments, no union negotiations, no reputational exposure. For the 24 people whose roles were not refilled, the story is different. They are not counted as unemployed. They are scattered. And the entry pathway into financial services customer operations has quietly closed.
| Role Category | AI Exposure | NZ Employment (est.) | Trend 2022–25 | Primary AI Threat | Est. Runway |
|---|---|---|---|---|---|
| Data Entry / Processing Clerks | Very High | ~8,200 | ↓ Declining Fast | OCR + LLM document processing | 1–3 years |
| Telemarketing / Outbound Sales | Very High | ~4,100 | ↓ Declining Fast | Conversational AI + voice synthesis | 2–4 years |
| Accounting / Bookkeeping Clerks | High | ~12,500 | ↓ Softening | AI accounting (Xero AI, Intuit AI) | 3–5 years |
| Junior Legal (discovery, research) | High | ~3,400 | ↓ Softening | LLM document analysis / case research | 2–5 years |
| Insurance Claims Processing | High | ~5,800 | ↓ Softening | Automated claims adjudication | 3–6 years |
| Journalism / Content Production | Medium | ~6,200 | ↓ Disrupted | Generative content at scale | 5–8 years |
| IT Support (Tier 1 Helpdesk) | Medium | ~9,100 | → Evolving | AI triage / self-resolution tools | 4–7 years |
| Retail / Customer Service | Medium | ~85,000 | → Mixed | Self-serve / AI kiosks (partial) | 5–10 years |
| Construction / Trades | Low | ~120,000 | → Resilient | Limited — physical dexterity required | 10+ years |
| Aged Care / Disability Support | Very Low | ~65,000 | ↑ Growing | Minimal — emotional intelligence required | 15+ years |
The NZ Government is caught in a structural contradiction: it is simultaneously mandating aggressive AI adoption across the public service while cutting the workforce that AI is supposed to help transition. The $2.4 billion public sector savings target has been explicitly linked by the Efficiency Minister to AI-enabled headcount reduction — conflating efficiency with austerity in ways that alarm public sector unions about impacts on the most vulnerable New Zealanders.
The planned reduction of 8,700 public sector roles — approximately 8% of the public service — assumes AI can absorb routine processing work before the staff who do it have left. The PSA and NZCTU counter that AI systems require significant human oversight, error correction, and exception handling, especially in high-stakes contexts (welfare eligibility, tax disputes, immigration decisions). Cutting staff before AI is proven reliable creates a gap where complex cases fall through.
Meanwhile, 81% of Kiwis support stronger AI regulation (Ipsos 2025) — yet the government's regulatory approach is explicitly "light touch," relying on existing employment and privacy law rather than AI-specific worker protections. This is the central political fault line in New Zealand's AI governance debate.
| Agency | Primary Function | AI Exposure | Planned FTE Cut (est.) | Key AI Application | Service Quality Risk |
|---|---|---|---|---|---|
| Inland Revenue (IRD) | Tax collection, compliance | High | ~750 FTE | Automated audit, query chatbot, returns processing | Medium |
| MBIE | Business services, employment | High | ~620 FTE | Permit processing, compliance monitoring | Medium |
| DIA (Births, Deaths, Passports) | Identity & citizenship services | High | ~340 FTE | Document processing, biometric ID verification | Lower — mostly routine processing |
| Ministry of Social Development | Welfare, employment support | Medium | ~980 FTE | Benefits eligibility screening, appointment triage | High — vulnerable clients at elevated risk |
| Ministry of Health | Health system oversight | Medium | ~420 FTE | Data analytics, clinical reporting automation | Medium |
| Immigration New Zealand | Visa, residency processing | High | ~380 FTE | Application pre-screening, document authenticity | High — errors have serious human consequences |
| ACC | Accident compensation | Medium | ~290 FTE | Claims triage, medical document processing | Medium — rehabilitation needs human judgement |
| Courts (Ministry of Justice) | Court administration | Medium | ~180 FTE | Document filing, scheduling, registry automation | Lower — judicial functions constitutionally protected |
| Kāinga Ora / Housing NZ | Social housing | Medium | ~220 FTE | Needs assessment triage, maintenance processing | High — housing vulnerable families |
“81% of Kiwis support stronger AI regulation — yet the government's approach has been described as 'light touch' by the very union that represents the workers most affected.”
— NZCTU / Ipsos NZ 2025The global policy response to AI-driven labour displacement is deeply fragmented, moving at radically different speeds. The EU has moved furthest, embedding AI employment protections into binding law through the EU AI Act (2024). The US remains in a legislative holding pattern. The UK is grappling with evidence that it is losing jobs to AI faster than any other G7 economy. And New Zealand — which typically takes regulatory cues from the UK — is watching from the sidelines with no equivalent legislation, no mandatory AI displacement reporting, and a regulatory posture that even sympathetic observers describe as inadequate.
The strategic opportunity for NZ: as a fast-follower, it can learn from others' early-mover mistakes and leapfrog to a more mature framework. The historical pattern in NZ labour policy suggests waiting until the problem becomes acute. The AI transition may be moving too fast for that luxury.
The goal is not to protect routine jobs — that argument is economically unwinnable. The goal is to ensure the transition does not strand a generation of young workers permanently below the broken rungs, and that productivity gains from AI are distributed equitably rather than captured entirely by capital.
This panel aggregates granular intelligence on skill demand shifts, emerging roles, and the AI adoption timeline in New Zealand. It is intended as an operational resource for policy-makers, career advisors, and employers navigating the transition. Data draws on MBIE vacancy surveys, Seek NZ postings analysis, LinkedIn Economic Graph NZ data, and the IDC/Deel 2025 workforce report. Australian proxy data is used and clearly labelled where NZ-specific data is unavailable.
| Emerging Role | Status in NZ | Median Salary (NZD) | Entry Requirements | Growth | NZ Demand |
|---|---|---|---|---|---|
| AI Implementation Consultant Advises orgs on AI adoption strategy and tooling | Established | $130–$190k | Tech/business + AI fluency | ↑ Strong | High — mostly filled by contractors |
| AI Prompt Engineer Designs and optimises LLM-driven workflows | Emerging | $90–$140k | Domain expertise + technical writing | ↑ Very Strong | Growing — early stage in NZ |
| AI Ethics & Governance Lead Responsible AI use, bias audits, compliance | Early Stage | $120–$180k | Law / philosophy / tech + policy | ↑ Accelerating | Low but rising — govt driving demand |
| Human-AI Collaboration Designer Designs workflows where humans and AI complement each other | Nascent | $95–$145k | UX + organisational psychology + AI literacy | ↑ Strong | Small but fast-growing in large enterprise |
| Data Quality Analyst AI training data accuracy, labelling, curation | Established | $70–$110k | Analytical background; can retrain from admin | ↑ Steady | Moderate — some offshoring |
| Automation Analyst Maps and automates business processes using AI tools | Established | $80–$130k | Business analysis + AI tool literacy | ↑ Strong | High — strong demand in finance + govt |
| AI-Augmented Legal Counsel Senior lawyers skilled in AI-assisted research and drafting | Emerging | $160–$280k | Admitted barrister/solicitor + AI proficiency | ↑ Growing | Growing — displacing junior roles as it scales |
| AI Trainer / Model Fine-tuner Human feedback for RLHF and model alignment | Early Stage | $55–$80k | Domain knowledge; accessible entry | → Mixed | Low — mostly offshore currently |