Is it really AI closing the door on junior jobs in NZ – or is it remote work?

The sharp decline in early-career hiring across four countries may have far less to do with ChatGPT than with the permanent shift to hybrid and remote work – and New Zealand has particular reason to pay attention.

The story has become familiar to every HR leader in New Zealand. Artificial intelligence is eliminating the entry-level job. It is doing the drafting, the data processing, the first-pass research that once kept graduate desks busy, and organisations no longer need to hire humans to do it. For two years, this has been the dominant explanation for one of the most troubling labour market trends of the post-pandemic era.

The evidence from New Zealand's own data is not encouraging. According to Stats NZ, the unemployment rate among 15- to 24-year-olds rose to 15.2% in the September 2025 quarter, up from 13.1% a year earlier — roughly three times the rate of the wider working-age population.The NEET rate — young people not in employment, education, or training — climbed to 13.8%. Junior office and administrative roles are increasingly disappearing from the labour market, and the usual explanation points to automation.

But a major new working paper — published in May 2026 by researchers at the University of Warwick, the London School of Economics, and Oxford's Ellison Institute of Technology — suggests that New Zealand HR leaders, like their counterparts across the English-speaking world, may be misreading the situation. The real culprit, the authors argue, may not be artificial intelligence at all. It may be the flexible work arrangements organisations worked so hard to embed during and after the pandemic.

The paper, The Broken Ladder: AI, Remote Work, and Early-Career Hiring, by Peter John Lambert and Yannick Schindler, draws on 243 million new hire records and 407 million online job postings across the United States, the United Kingdom, Canada, and Australia, spanning 2017 to 2025. New Zealand is not in the dataset, but the pattern it reveals is directly relevant.

When researchers measure the effect of generative AI exposure and working-from-home exposure on junior hiring separately, both appear powerful. A two-standard-deviation increase in either exposure predicts roughly a 4–5 percentage point fall in the junior share of new hires by 2025. Both tell a convincing standalone story.

The problem is that they are measuring largely the same thing. The WFH exposure index and the GenAI exposure index used in the study have a Spearman rank correlation of 0.77 across 683 occupations. Software developers, accountants, and management consultants sit near the top of both rankings. Electricians, construction labourers, and janitors sit near the bottom. The occupations most amenable to remote work are, almost without exception, also the occupations most exposed to AI automation.

When both variables enter the model simultaneously, the picture shifts dramatically. The WFH effect holds firm. The AI effect collapses — often becoming statistically indistinguishable from zero. This is not a minor methodological footnote. It is a direct challenge to the prevailing explanation for what has happened to early-career hiring across the four economies studied.

New Zealand occupies an unusual position in this debate. Roughly 70% of New Zealand's exports come from agriculture, horticulture, seafood, and forestry, while leading employment sectors include aged care, physiotherapy, plumbing, childcare, and early childhood education — roles requiring physical dexterity, sensory judgement, and human empathy that AI cannot yet credibly replicate. New Zealand's economic base arguably offers some structural insulation from AI disruption in ways that are not available to more services-intensive economies.

But that structural insulation does not extend to the knowledge-intensive, white-collar occupations where New Zealand's junior hiring decline is concentrated. According to Stats NZ, 34% of employed New Zealanders worked from home either part-time or regularly as of 2025 — a figure that has held broadly stable since 2022 and shows no signs of returning to pre-pandemic levels. It is precisely in these remote-capable, knowledge-intensive roles that both WFH adoption and GenAI exposure are highest, and where the junior hiring decline is most acute.

New Zealand also faces a compounding problem that makes the junior pipeline especially fragile. Companies are losing experienced talent across the Tasman and to the United Kingdom, forcing employers to rethink how they hire. When senior and mid-level workers leave, the internal career ladder that junior hires are meant to climb becomes structurally shorter. The incentive to hire at the bottom of that ladder weakens further.

The Lambert-Schindler paper's theoretical framework is straightforward, and its intuition will resonate with any HR leader who has tried to onboard graduates into a distributed team. Firms hire junior workers not just for their immediate output, but as an investment in the workforce they will become. That investment requires mechanisms for knowledge transfer: proximity to senior colleagues, informal feedback, the gradual accumulation of tacit skills that cannot be conveyed in a training module.

Remote and hybrid arrangements raise the cost of all of that. The research literature cited in the paper is consistent. Studies of collaboration networks found that firm-wide remote work made them more siloed. Research on young software engineers found that in-person time delivered the largest productivity benefits for the youngest workers. Research published in The Quarterly Journal of Economics in 2026 found that proximity to co-workers significantly increases the quality and frequency of feedback for early-career employees.

The model in the paper formalises the mechanism: as supervision costs rise and on-the-job learning rates fall, the rational firm tilts its hiring toward experienced workers who have already built their capabilities — and who need less organisational scaffolding. That tilt, replicated across thousands of firms, produces the aggregate junior hiring decline visible in the data.

None of this is an argument against flexible work. It is an argument that flexible work, when applied without deliberate adjustment to how early-career talent is developed, makes the return on junior investment harder to see — and that firms respond by hiring fewer juniors.

It would be a mistake to read the Lambert-Schindler paper as clearing AI of any labour market impact. There are genuine warnings against replacing entry-level employees with AI, with researchers noting that doing so risks cutting off the talent pipeline and ultimately creating a shortage of skilled mid- and senior-level professionals who never had the opportunity to build foundational experience.

Reports from New Zealand show that companies are prioritising AI fluency above other skills and cutting analyst and consultant roles as AI deployment accelerates. These are real effects on real occupations. The paper does not dispute that AI is reshaping what junior workers do. What it challenges is whether AI is the primary reason firms are choosing not to hire junior workers at all.

The robustness of the WFH finding is notable. Across alternative exposure measures, leave-one-out country and occupation tests, non-parametric controls, and measurement-error simulations, the WFH effect holds. A confounder explaining only 1–3% of residual variation in both treatment and outcome would be enough to eliminate the AI coefficient entirely. Eliminating the WFH coefficient would require a confounder five times as powerful.

The paper also presents a compelling direct test. When the researchers replace predicted exposure measures with actual observed WFH adoption — tracking firms and regions that explicitly offered remote and hybrid roles in 2021–22 — they still find a robust negative effect on junior hiring in the years that followed. The causal story survives contact with real adoption data.

The policy implications are more optimistic than the AI-displacement story allows — but only if organisations are willing to act on them. If AI were the primary driver of declining junior hiring, remedies would have to be systemic: government wage subsidies, training levies, industry-wide graduate quota arrangements. If WFH is the primary driver, the remedies lie within the reach of every HR function in the country.

For New Zealand organisations, the immediate practical questions are diagnostic. Does your junior hiring decline correlate most closely with the business units that went hybrid earliest — or with those deploying AI most aggressively? Have you rebuilt your onboarding and mentoring infrastructure to compensate for the loss of incidental proximity that hybrid work removes? Does your hybrid policy apply the same framework to a third-year analyst as to a graduate in their first six months?

The deeper problem is not just whether young New Zealanders can find jobs. It is whether the labour market still offers them a way in — and the evidence from Lambert and Schindler suggests that if it does not, the reason may be organisational rather than technological. That is a harder diagnosis than blaming AI. It is also one that organisations have the capacity to address.

New Zealand's talent shortage is already acute, worsened by outward migration of experienced workers. The pipeline that feeds tomorrow's experienced talent is today's cohort of junior hires. If organisations stop hiring at the bottom of the career ladder because hybrid work has made that investment feel too uncertain, they will find the top of the ladder emptying faster than they can fill it.

The broken ladder is a management problem. And management problems, unlike technological disruption, can be fixed.