Is AI really killing Singapore's entry-level jobs – or is hybrid work the hidden driver?

The numbers from Singapore's own graduate employment surveys have been quietly troubling for two years. According to the Ministry of Education's Joint Autonomous Universities Graduate Employment Survey, just 74.4% of 2025 university graduates in the labour force secured a full-time job — down from 87.5% of their 2022 counterparts. The proportion unemployed and actively searching rose to 10.5%, compared with 3.6% in 2022. The median gross monthly salary has held flat at S$4,500 for two consecutive years.

The dominant explanation across boardrooms and business schools has been a single word: AI. Generative tools are doing the analysis, the drafting, the data work that fresh graduates once performed, and organisations are concluding they no longer need to hire at the bottom of the career ladder to get it done.

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 — challenges that account in ways Singapore HR leaders cannot afford to ignore. Drawing on 243 million new hire records and 407 million job postings across the United States, the United Kingdom, Canada, and Australia between 2017 and 2025, the study finds that when the effects of AI exposure and working-from-home exposure are properly separated, the WFH effect holds firm — and the AI effect largely disappears.

The paper, The Broken Ladder: AI, Remote Work, and Early-Career Hiring, by Peter John Lambert and Yannick Schindler, makes a methodological argument with significant practical consequences. Both GenAI exposure and WFH exposure, measured at the occupation level, appear to be powerful predictors of junior hiring decline when analysed in isolation. A two-standard-deviation increase in either predicts a fall of roughly 4–5 percentage points in the junior share of new hires by 2025.

The reason both effects look similar is that they are measuring largely the same underlying occupational characteristic. The WFH exposure index and the GenAI exposure index have a Spearman rank correlation of 0.77 across 683 occupations. Software developers, data scientists, accountants, and management consultants sit near the top of both rankings. Construction workers, care staff, and technicians sit near the bottom. In Singapore's white-collar, knowledge-intensive employment hub, this correlation is especially high — the city-state's economic specialisation means its most significant occupational categories are concentrated precisely where both exposures are greatest.

When both variables enter the model simultaneously, the WFH coefficient remains stable and statistically significant. The AI coefficient attenuates sharply and often becomes indistinguishable from zero. The authors subject this finding to extensive robustness testing — alternative exposure measures, leave-one-out country and occupation tests, non-parametric controls, measurement-error simulations — and the WFH-dominant pattern holds across all of them.

Singapore is a particularly instructive case through which to read this research. The government has been a determined promoter of flexible work arrangements. The Tripartite Guidelines on Flexible Work Arrangement Requests, which came into effect in December 2024, require employers to formally consider employee requests for flexible arrangements, including work-from-home. The policy was designed to improve work-life balance and labour force participation — worthy objectives that carry unintended side effects if the Lambert-Schindler thesis is correct.

LinkedIn data has previously shown that employers in Singapore continue to prioritise onsite work for entry-level positions — a finding consistent with the paper's logic. If firms believe in-person presence is important for junior workers, the natural response to formalised hybrid norms is not to bring junior staff back to the office more often, but to avoid hiring junior staff whose development will be difficult to manage under distributed arrangements.

Indeed data shows that the share of job postings in Singapore mentioning remote work has been rising, even as overall job posting volumes have declined for consecutive months. The occupations gaining remote mentions — finance, professional services, technology — are precisely those where the junior hiring squeeze is most pronounced. If the Lambert-Schindler paper's mechanism is operating in Singapore, rising WFH availability in these sectors would be expected to suppress junior hiring even as firms attribute their decisions to AI capability.

The paper's conceptual framework is intuitive to anyone who has managed an early-career graduate in a distributed team. Firms hire junior workers not just for the tasks they complete today, but as an investment in the experienced professionals they will become. That investment depends on mechanisms for knowledge transfer that remote work fundamentally disrupts: informal feedback in shared physical spaces, incidental observation of how senior colleagues handle complex problems, the gradual accumulation of tacit judgement that cannot be scheduled into a video call.

As supervision costs rise and on-the-job learning rates fall — both empirically documented effects of remote work — the rational firm tilts its hiring toward workers who have already accumulated those capabilities and who need less organisational support to be productive from their first week. That tilt, multiplied across thousands of firms in Singapore's compact and interconnected business community, produces the hiring pattern now being observed.

Research cited in the paper found that collaboration networks became more siloed under firm-wide remote work. A study of software engineers found that in-person time delivered the largest productivity benefits for the youngest workers. Research published in 2026 found that proximity to co-workers significantly increases the quality and frequency of feedback for early-career employees. These are not arguments against flexible work: they are arguments for designing it carefully, with specific provision for early-career populations.

The SUTD's recent analysis of Singapore's graduate employment situation is illuminating in this context. Entry-level vacancies, while declining from 34,600 in end-2024 to 32,500 in end-2025, remain high in absolute terms. Yet full-time employment rates for graduates continue to fall. The paper notes that junior roles across sectors now require data analysis skills, familiarity with AI tools, and soft skills of the kind typically associated with third-year associates rather than fresh graduates — precisely the pattern the Lambert-Schindler paper would predict if firms are raising the bar for junior hires rather than eliminating junior roles entirely.

LinkedIn data shows that more than one in ten workers hired in 2024 have job titles that did not exist in 2000, and 78% of HR professionals in APAC say their organisation is prioritising upskilling initiatives in 2025. This is consistent with a labour market in transition, but not necessarily one in collapse. What appears to be happening — consistent with the Lambert-Schindler hypothesis — is that firms are demanding more from the junior workers they do hire, because those workers will need to operate productively under hybrid conditions from the outset.

A Harvard working paper by Seyed Hosseini and Guy Lichtinger found that at firms that adopted generative AI, junior employment declined by approximately 9% after six quarters relative to non-adopting firms — an AI effect that is real and documented. The Lambert-Schindler contribution is not to deny this but to ask whether, in the absence of controls for WFH adoption, AI-exposure designs may be attributing some or all of the WFH effect to AI.

The distinction matters enormously for how HR leaders respond. If AI is the primary driver of declining junior hiring, the remedies are largely outside individual organisations' control — government policy, industry-level intervention, structural economic adjustment. If WFH is the primary driver, the remedies are precisely within the remit of HR: onboarding design, structured mentoring, deliberate use of in-person time for early-career cohorts, and hybrid policies calibrated to tenure and development stage rather than applied uniformly across experience levels.

Singapore's HR community has consistently been at the forefront of progressive workforce practice in the region. The same creativity that built world-class SkillsFuture programmes and sophisticated succession pipelines can be applied to this problem. But it requires first identifying the problem correctly.

The practical audit for Singapore HR leaders begins with a simple correlation: does your organisation's junior hiring decline track most closely with the business units and job families that adopted hybrid and remote work earliest — or with those deploying AI tools most aggressively? For most organisations in Singapore's professional services, finance, and technology sectors, the honest answer to that question may be more illuminating than any AI benchmarking exercise.

The career ladder into Singapore's most sought-after professions is narrowing. The Lambert-Schindler paper offers reason to think the cause is not inevitable technological disruption, but a manageable organisational challenge. For a city-state whose competitive advantage rests on talent development, that distinction is not academic. It is strategic.