The number to hold: six in ten, split down the middle
The most cited figure in the AI-jobs debate — “40 percent of jobs worldwide are exposed to AI” — got a New Zealand-specific answer this year, and it is bigger than the global number. An IMF staff research paper on New Zealand, published as part of the IMF’s 2026 Article IV consultation materials, estimates that about 60 percent of New Zealand workers hold jobs in high-exposure occupations, in line with Australia, the UK and the US.
But the paper’s second half is the part that matters for anyone actually managing a career through this. Exposure, in the IMF’s framework, is not fate: of that 60 percent, the roles split roughly evenly between occupations where AI tends to complement the worker and occupations where it risks displacement. Only around one-third of all workers sit in the genuinely vulnerable band — high exposure combined with low complementarity — and who they are complicates the usual narrative.
Who actually holds the risk
The vulnerable third, per the paper, is disproportionately female and concentrated in the middle of the income distribution — not the bottom, not the top. Women’s employment skews toward high-exposure occupations (the paper reports higher exposure shares for female workers than male across their occupational mapping), and because women’s exposure spans both complementary and displacement-prone roles, they face what the authors call dual exposure: stronger augmentation potential in some occupations, elevated displacement risk in others. That mirrors the IMF’s cross-country finding that women are simultaneously more likely to benefit from AI and more vulnerable to displacement, with lower-earning women in substitution-prone jobs carrying the most risk.
Meanwhile, exposure rises with income — and so does the potential for AI to complement rather than replace. The paper’s modelling finds the augmentation upside strongest in the upper-middle of the earnings distribution, which is one reason the “high exposure equals high risk” framing keeps failing the data.
The confirming signals in NZ’s own labour market
Nothing in the official NZ numbers contradicts the IMF picture, and several pieces of it have since moved in directions worth tracking:
- Job-ads demand keeps rising. SEEK NZ’s September employment report, released mid-October, showed job ads up 7.8 percent year-on-year, with AI-skills mentions at a record 4.2 percent of listings — and employers naming AI at nearly double Australia’s rate. The record inside that report was the fastest-growing asks being governance and supervision, not basic prompting.
- The wage channel is real but uneven. This year’s international pay data — including the Indeed analysis of AI-exposed pay premiums — found AI skills commanding premiums in exposed occupations, consistent with the IMF’s complementarity channel rather than its displacement one.
- Preparedness is a strength, not a slogan. The paper ranks New Zealand’s AI preparedness among the best in the world via the IMF’s AI Preparedness Index, while flagging the structural bottlenecks — low R&D intensity notably, high electricity prices, shallow capital markets — that could slow diffusion from frontier firms to everyone else. It also finds NZ demand tilted toward applying AI rather than building it: the share of postings requiring both developer and user skills has risen from about 24 percent in 2022 to about 34 percent in 2026, which the authors read as employers valuing workers who can both apply and adapt AI tools.
What the IMF model actually says could go either way
The paper’s economic simulations are the constructive part: under favourable conditions — including a properly managed transition for displaced workers — AI could raise NZ output materially over the coming decade. The labour-market risk it flags is real but conditional: around one-third of workers in high-risk occupations, with the outcome depending on whether productivity gains generate new labour demand and whether affected workers can move into complementary roles.
Its prescription is unglamorous and specific: active labour-market policies, accessible retraining and lifelong-learning systems, broad diffusion of AI-user skills, and social protection that cushions longer transitions. That is the same architecture showing up in employer-led examples offshore — including the short-course, employer-funded pathways that US infrastructure projects are funding — and it tracks what New Zealand’s own officials have been warned about since Treasury’s earlier exposure work.
For women — the group the analysis singles out — the practical reading is sharper: the paper implies the displacement risk concentrates in mid-income clerical and administrative-adjacent work, exactly the band where Official UK data this month found AI’s heaviest reported effects, even as headcount effects stayed modest there.
🔍 THE BOTTOM LINE
- An IMF staff paper estimates about 60 percent of New Zealand workers are in AI high-exposure occupations — above the 40 percent global figure, in line with other advanced economies.
- Exposure splits roughly evenly between augmentation-type and displacement-risk roles; only about one-third of workers overall sit in the high-risk, low-complementarity band.
- The vulnerable third is disproportionately female and mid-income — not concentrated among the lowest paid, contrary to common framing.
- NZ’s preparedness ranks among the best in the world; the simulation upside is a material output gain this decade if the transition is managed via retraining and lifelong-learning systems.
- Current NZ data — job ads up 7.8 percent annually, AI-skills mentions at a record 4.2 percent of listings — runs on the complementarity side of the ledger so far.
❓ FAQ
Are 60 percent of NZ jobs really at risk from AI? No — 60 percent are in high-exposure occupations, and the IMF’s framework splits those between roles where AI complements the worker and roles where it may displace them. The paper puts the genuinely vulnerable band — high exposure plus low complementarity — at around one-third of workers. Exposure is a measure of contact with the technology, not of job loss.
Why are women described as more exposed? The paper’s occupational mapping puts a larger share of female employment in high-exposure roles, and spans both complementary and displacement-prone occupations — so women face both stronger potential upside and elevated risk in specific bands. Its cross-country reference work identifies low-earning women in substitution-prone jobs as the most at-risk group.
What does this mean for choosing a career path? The paper’s own data directions — rising AI-skills demand, premiums for AI-adjacent skills, strongest augmentation in the upper-middle of the income distribution — favour roles that sit next to AI systems (supervising, governing, deploying) over roles AI does the core of. New Zealand’s job-ads mix is trending that way.
What does the IMF say New Zealand should do? Accessible retraining and lifelong-learning systems, active labour-market policies, broad diffusion of AI-user skills, and social protection for longer transitions — plus structural fixes (R&D incentives, deeper capital markets) so AI gains diffuse beyond frontier firms.
— CJ Murden, editor of Singularity.Kiwi. Former digital technologies teacher, author of AI-focused books. Writing with a New Zealand focus.
