In five Arab countries with available data, administrative support jobs represent between 1.0% and 6.9% of youth employment, without specific breakdown of the formal sector. According to the ILO, the youth unemployment rate was 28.0% in 2023 and was projected at 28.6% in 2024, more than double the global average of 13%. The ILO estimates that approximately 14.6% of jobs in the 12 Arab states studied present high potential for augmentation by generative AI and 2.2% potential for automation. An entire generation awaits access to these positions at a time when these technological transformations are accelerating.
The Essentials
- The ILO projected youth unemployment of 28.6% in Arab states for 2024, more than double the global average.
- Generative AI particularly exposes clerical tasks, but exposure does not allow us to conclude that formal entry pathways will disappear as a priority. Projections of AI’s effects on employment extend to 2035, but exposure estimates describe the observed employment structure and distinguish between augmentation and automation.
- Nearly half of organizations in the Middle East report lacking the technological talent and capabilities necessary for large-scale deployment. A skills deficit may limit the benefits of AI, but net employment effects depend on policies, investments, and adoption modalities.
- Economist Axelle Arquié identifies a double shock: employment and taxation. Gulf states historically have low direct taxation, but generally have substantial budgetary and financial buffers related to hydrocarbons, while Mashreq and Maghreb countries have more developed fiscal systems but constrained budgets.
- Training undertaken after a shock can still contribute to adaptation, retraining, and redeployment, even if early preparation can reduce transition costs.
The Jobs That AI Erases First Are Not the Most Visible
Public institutions will have a limited window to act before an entire generation is permanently excluded from the labor market.
A junior accountant in a Jordanian bank, a data entry officer in an Egyptian ministry, an administrative assistant in a Saudi SME: these professions share one characteristic. They are routine, codifiable, and they constitute the first rung of a formal career ladder for higher education graduates. These are precisely the jobs that generative AI systems and document workflow automation absorb most easily.
The ILO, in its analysis published in 2026 on AI transformation of Arab labor markets, identifies administrative and clerical tasks as particularly exposed to generative AI. The public sector remains a dominant employer in several Gulf and Mashreq countries. When these positions disappear or cease to be created, pathways to formal labor market access narrow.
The mechanism can be insidious: some organizations may slow recruitment for these positions without eliminating existing posts, reducing access opportunities for new graduates. At 28.6% youth unemployment, the Arab region did not need this additional tightening.
The Skills Deficit for AI Adoption That We Fear
The Arab region faces risks of uneven transformation and preparation gaps. According to Deloitte Middle East, nearly half of organizations lack the technological talent and capabilities for large-scale deployment.
Companies that automate without the skills to optimize their tools achieve partial productivity gains. Those who wish to go further in AI adoption cannot find the right profiles. Meanwhile, graduates leave universities that still largely train for the professions that AI is absorbing.
This gap between training supply and skills demand is not unique to the Arab region. Automated warehouses in Europe also lack qualified personnel to operate them. But the Arab region presents a particularly unfavorable combination: already high youth unemployment, limited private sector diversification in several countries, and educational systems whose reform is slow in the face of AI’s spread.
The Double Shock That Axelle Arquié Identified, and Why It Is Amplified in the Middle East
Axelle Arquié, in The Double Shock of AI: Employment and Taxation (2025), constructs a two-part argument. The first shock is direct: AI destroys or transforms jobs, now affecting qualified functions that previous waves of automation had spared. The second shock is fiscal: if AI captures the value created by labor and transfers it to capital, states lose a portion of the tax base on which public services and social protection financing rests.
In European economies, these two shocks arrive on systems that have buffers. Income tax, social contributions, active employment policies form a safety net. These nets may be insufficient, and Arquié acknowledges this, but they exist. In several Arab countries, the model is different. Gulf economies function with very low direct taxation, historically compensated by oil revenues redistributed through public employment.
Some, however, have substantial budgetary and financial buffers related to hydrocarbons.
Mashreq and Maghreb countries have more developed fiscal systems but constrained budgets. Existing institutional architectures will face shocks of varying magnitude depending on available response capacities.
Arquié’s thesis on the need for intelligent AI regulation, capable of preserving fiscal bases without falling into technological protectionism, poses with particular urgency here. A country like Saudi Arabia, which aims to reduce its hydrocarbon dependence through Vision 2030, is investing massively in digitalization. But digitizing without redistributing technological gains amounts to substituting oil rents with algorithmic rents, with the same concentration problem and the same fragility over time. This reading aligns with that of economists like Daron Acemoglu and Simon Johnson, who insist on the question of who captures the gains from technological progress. The Arab region provides a full-scale test of their thesis: the capital-labor divide is playing out in accelerated fashion.
The Lesson of 1990s South Korea on the Timing of Investment
1990s South Korea pursued its climb toward high-technology industries, supported by prior investments in skills. The Korean lesson is often misread: what matters is the timing of public investment in vocational training and the prior existence of institutions capable of adapting rapidly.
South Korea already had a long-standing vocational training system and introduced employment insurance in 1995 supporting training and job search. Vocational training institutions existed, curricula evolved, partnerships with chaebols for direct job placement worked. When the shock came, the system had absorption capacity. Displaced workers could find retraining toward qualified jobs thanks to the prior existence of these pathways and their credibility with employers.
The Arab region finds itself in a different situation: retraining institutions are still being built in the face of rapid technological changes. Training data analysts, AI integrators, cybersecurity specialists takes time. Building the credibility of these pathways with employers and families takes even longer. Programs announced, from digital training initiatives in Saudi Arabia within Vision 2030 to Egypt’s efforts in technological skills, move in the right direction.
But their deployment encounters rapid technological changes.
What Will Determine Whether a Generation Is Lost by 2030-2035
The question posed by the ILO invites examination of decisions made in Arab capitals. The directions taken will have significant bearing on the years that follow. Two trajectories are taking shape, and they depend on a small number of variables.
In the first trajectory, governments in the region accelerate reform of their training systems in partnership with the private sector, on the model of intensive technical bootcamps that have produced results in Estonia and India. These training programs do not replace university but create alternative pathways to the labor market, in six months rather than four years. The condition is that the private sector validates these pathways at hiring, which requires certification and standardization work that states must drive. If this dynamic were to take hold in two or three countries, with measurable results on insertion rates at twenty-four months, it could spread. Signals to watch are absorption rates for these programs and the capacity of regional technology companies to recruit locally rather than import skills.
In the second trajectory, the rapid spread of AI would demand considerable institutional response capacities. Higher education graduates could encounter difficulties finding jobs matching their qualifications, and migration flows could experience changes related to these labor market tensions. This trajectory would present risks comparable to those observed during past trade shocks in industrialized countries, but the analogy with the relocations of the 1980s-2000s remains hypothetical, as the mechanisms and demographic composition of the Arab region differ significantly. The difference is that the Arab region overall has a younger population, which changes the scale and horizons of the transition, with distinct risks and opportunities.
What distinguishes the two scenarios depends more on institutional architecture than on financial resources alone. Gulf countries have the means to invest. Mashreq countries sometimes have university infrastructure. What may be missing in both cases is the capacity to coordinate education ministries, private companies, and financing institutions around shared insertion objectives. Investment in R&D without visas to attract talent fails everywhere: the Arab region teaches that investment in training without employers to validate skills fails just as much.
The most promising bet may not be training AI developers, but training AI integrators in sectors where the region has massive unmet needs: healthcare, logistics, water management, energy transition. These domains combine strong local demand, jobs that cannot be easily relocated, and AI use as a tool of a profession rather than as its substitute. This is a model of productive coexistence with technology, and several experiments are moving in this direction in the UAE and Jordan. Their scaling up will condition what the Arab region will have to show the rest of the world by 2035.
Sources
- ILO, AI-driven transformation puts jobs, inequality and skills at crossroads in Arab region (2026)
- Axelle Arquié, The Double Shock of AI: Employment and Taxation (2025), Alternatives Économiques, alternatives-economiques.fr
- Deloitte Middle East, report on AI skills in the region (cited via Qatar Tribune, 2026; exact URL not available)
