The conventional reading of the International Labour Organization’s latest youth employment report is straightforward: AI is coming for young workers, and the numbers are getting worse. It is a compelling headline, and it is largely wrong — or at least, it misidentifies the mechanism doing the most damage right now.
The ILO’s July 2025 report shows global youth unemployment climbed to 12.4% last year, up from 12.3% in 2023, representing roughly 67 million people aged 15 to 24 who cannot find work. But buried in the data is a more structurally significant finding: the primary engine of youth job destruction today is not algorithmic displacement — it is the accelerating hollowing-out of middle-skilled roles that have historically served as the primary on-ramp into the formal labour market.
The Market Today
The global youth labour market — broadly defined as employment conditions for workers aged 15 to 24 — is a segment of enormous economic consequence. The ILO estimates that more than 257 million young people worldwide currently sit in the NEET category: not in employment, education, or training. That figure, representing approximately 20% of the global youth population, edged upward in 2025, reversing improvements that had emerged in the aftermath of the COVID-19 pandemic.
The deterioration was geographically broad. Youth unemployment rose between 2023 and 2025 in eight of the world’s eleven subregions, spanning high-income, lower-middle-income, and low-income economies alike. The Arab States recorded the highest youth unemployment rate at 26.2%, followed by Northern Africa at 22.6%. What is analytically striking — and frequently underreported — is the pace of deterioration in wealthier economies. Youth unemployment in North America rose to 9.8% in 2025 from 8.3% in 2023, a two-percentage-point swing in two years. Across Northern, Southern, and Western Europe, the rate stood at 15%.
These are not marginal shifts. They represent millions of additional young people cycling through extended job searches, downgraded employment, or outright withdrawal from the labour force. The structural implications for consumer spending, household formation, and long-run productivity are material, even if they accrue slowly enough to escape quarterly earnings calls.
The Major Players
The International Labour Organization
The ILO, the United Nations’ labour agency, is the authoritative source for these figures and the primary institution shaping policy responses. Its annual youth employment reports set the analytical frame for governments, development banks, and multinational employers. The agency is not a market actor in the commercial sense, but its threat assessments carry significant weight with policymakers in OECD economies and multilateral lenders.
High-Income Country Governments
Advanced economies are simultaneously the largest funders of workforce retraining programmes and the jurisdictions where the middle-skill hollowing is most advanced. Governments in North America and Europe face a structurally awkward position: the job categories most exposed to both automation and offshoring are precisely the ones they have historically relied upon to absorb secondary school and university graduates who do not pursue professional or technical careers.
Technology Sector Employers
Big Tech firms and AI developers occupy a dual role. They are the agents of the structural change driving middle-skill displacement, and they are simultaneously among the fastest-growing employers of high-skilled young workers in STEM and data-intensive fields. The ILO notes that jobs in science, engineering, healthcare, and information technology continue to expand in most countries — but these roles demand qualifications and experience that a significant share of young labour-market entrants do not yet possess.
Informal Economy Employers in Developing Nations
In low- and lower-middle-income countries, the relevant “market” is largely informal. Nearly nine in ten young workers in these economies remain in informal employment, without adequate labour protections or social safety nets. For these workers, the choice is not between a middle-skill formal job and an AI-assisted gig — it is between insecure informal work and nothing at all.
The Assumed Story
The dominant media narrative frames the youth jobs crisis as a preview of an AI displacement wave. Under this reading, chatbots, automation, and large language models are already eroding entry-level white-collar work — the sort of administrative, clerical, and analytical tasks that recent graduates have historically absorbed — and the ILO figures are early evidence of this trend becoming visible in macro data.
This framing is not entirely without foundation. The ILO does flag AI exposure as a meaningful forward risk: it estimates that 6.1% of jobs currently held by workers aged 15 to 29 are in occupations most exposed to AI-related changes. If just 10% of those positions disappeared entirely, approximately 5.6 million young workers — concentrated in high-income countries — could face unemployment, forced career changes, or exit the labour force altogether.
But that is a conditional, second-order scenario. The ILO itself uses careful hedging language, noting that “the long-term impact of AI remains uncertain.” What is already happening, by contrast, is considerably more prosaic and considerably more damaging at scale.
Where Capital Is Moving
Investment flows in the labour market — both private and public — tell a revealing story about where structural adjustment is and is not occurring. Venture and corporate capital is pouring into AI-driven productivity tools, particularly those targeting white-collar workflows: document processing, customer service, code generation, and data analysis. Many of these tools are explicitly positioned to reduce headcount in the clerical, administrative, and junior analytical roles that the ILO identifies as the traditional entry points for young workers.
At the same time, public investment in workforce development has not kept pace with the speed of structural change. Research increasingly suggests that AI’s most immediate labour market effect is not headline job loss but wage suppression — employers using the credible threat of automation to resist pay increases rather than immediately replacing workers. For young people entering the labour market with high debt loads and limited negotiating leverage, this dynamic is economically corrosive even when their jobs nominally persist.
The concentration of high-skill job growth in sectors like healthcare IT, engineering, and data science also raises a capital-allocation question: who is financing the education and training pipelines that would allow a meaningful share of today’s NEET population to access those roles? The gap between where labour demand is growing and where current educational systems are producing graduates represents both a market failure and, potentially, a significant investment opportunity for the right platforms.
There is an underappreciated geographic irony embedded in the ILO data. The AI exposure risk — the 5.6 million conditional job-loss estimate — is concentrated in high-income countries, where young workers at least have access to retraining resources, social safety nets, and formal labour protections. The structural crisis that is already happening — near-universal informal employment — is most acute in low- and lower-middle-income countries, where the concept of AI displacement is almost entirely irrelevant to daily economic reality. The global policy conversation, dominated by wealthy-country think tanks and media, is arguably optimised for the smaller and more speculative of the two problems.
The Strongest Counterargument
The most serious objection to framing the middle-skill hollowing as the primary driver — rather than AI — comes from labour economists who argue that the two phenomena are not separable. Automation of routine cognitive tasks, a process that predates large language models, has been dismantling middle-skill jobs for at least two decades. The argument, associated with the academic work of economists such as Daron Acemoglu and David Autor at MIT, holds that we are already deep inside a multi-decade “task displacement” cycle, and that modern AI tools represent an acceleration of the same force, not a new one. Under this reading, distinguishing between “AI displacement” and “middle-skill hollowing” is a false dichotomy — they are the same structural shift, just at different speeds.
This is a genuinely strong objection, and it has empirical support. But it does not necessarily weaken the contrarian frame — it sharpens it. If AI is best understood as an accelerant of a longer-running structural trend rather than a new cause, the policy implication changes significantly. The problem is not how to prepare for an AI disruption that is coming; it is how to address a structural dislocation that has been accumulating for a generation and is now moving fast enough to show up clearly in ILO macro data. That reframing suggests that workforce investment, not technology governance, is the most urgent lever. More than 200 economists and AI leaders have signed urgent warnings about job displacement, but the debate often centres on future-tense governance rather than present-tense structural investment.
Financial and Strategic Implications
For incumbents in sectors historically reliant on large pools of entry-level labour — retail, financial services administration, call centres, back-office processing — the ILO data suggests that the transition away from middle-skill hiring is already well advanced. The macro unemployment numbers are, in part, the trailing indicator of hiring decisions made over the previous several years.
For technology companies building AI productivity tools, the data presents both a commercial and a reputational consideration. Commercially, the continued expansion of NEET populations in high-income markets suggests strong latent demand for reskilling platforms, credentialing services, and AI-assisted career navigation tools. Reputationally, the growing social backlash against AI adoption is increasingly linked, in public perception, to exactly the kind of job market deterioration the ILO is documenting — even if the causal chain is more complex than headlines suggest.
For investors, the picture is nuanced. The sectors showing employment growth — healthcare, engineering, data infrastructure — are attracting capital, but they are also the sectors most insulated from the youth unemployment problem because they require advanced credentials. The more interesting investment question may be in the intermediary layer: the platforms, bootcamps, apprenticeship programmes, and credentialing bodies that could bridge the gap between where young workers are and where formal labour demand is growing. Census data already shows that AI adoption is concentrated in a subset of large firms, suggesting that the competitive advantage from early AI deployment — and the associated middle-skill displacement — is not uniformly distributed.
Risk Factors
Several dynamics could complicate or accelerate this thesis. First, if AI-driven productivity gains translate into stronger economic growth than current forecasts suggest, the overall demand for labour could expand in ways that offset middle-skill displacement — the canonical “comparative advantage” argument for technology-driven growth. The ILO report itself notes that high-skilled job categories are expanding; the question is whether that expansion is fast enough and accessible enough to absorb displaced workers across education and income levels.
Second, geopolitical fragmentation poses an independent risk. The ILO explicitly cites geopolitical tensions alongside weaker economic growth as structural headwinds for 2025. Supply chain restructuring, trade policy uncertainty, and regional conflict all reduce the cross-border investment flows that historically generate formal employment in developing economies — precisely the markets where the youth unemployment problem is simultaneously most severe and least likely to be addressed by AI governance frameworks designed in Washington or Brussels.
Third, the NEET figure deserves particular attention as a leading indicator. At 20% of global youth — 257 million people — the share of young people entirely outside employment, education, or training is not just a labour market problem. It is a social stability risk, a public health pressure, and a long-run fiscal liability for governments that will eventually bear the cost of extended workforce exclusion in the form of reduced tax revenue and elevated social spending. The ILO’s warning here is less about any single economic cycle and more about the compounding of structural disadvantage over time.
The Prediction
Within the next three years, the policy debate around youth employment will be forced to shift its primary focus from AI governance to structural workforce investment — not because AI risk recedes, but because the middle-skill hollowing already in the data will become impossible to attribute primarily to future technology. Governments in North America and Europe that fail to make that pivot will see youth unemployment rates stabilise above pre-pandemic highs. The signal that this prediction is wrong: a sustained, broad-based increase in formal middle-skill job creation that outpaces the current rate of role elimination — something no major labour market forecast currently projects.
