Large-scale automation has not yet reached Sub-Saharan Africa, but the window to build graduate capability is narrower than it looks.
By Dr Kelechi Ekuma, Senior Lecturer at GDI
More than 22 million Africans enter the labour market every year, and more of them hold a degree than at any point in the region’s history. Enrolment in higher education has expanded steadily for two decades. By standard measures of educational progress, Sub-Saharan Africa is doing what it was told.
And yet fewer than one in twenty young Africans hold advanced digital skills. Ninety-eight per cent of under-18s leave school without basic competencies in science, technology, engineering and mathematics, restricting their potential to enter professionalized careers in innovative sectors. It is no surprise, therefore, that young people not in employment, education or training (NEET) account for 20.4 per cent of the region’s youth. Participation has risen. Preparation has not.
Our recently published paper, The Persistent Capability Gap: Higher Education, Innovation Systems and Technological Unemployment in Sub-Saharan Africa, argues that this gap is not a puzzle but a predictable result, and that the policy debate has been asking the wrong question. The issue is not how many young Africans get into university. It’s how they can use higher education to pursue lucrative and fulfilling careers.
First, an honest caveat: the robots are not here yet
Any argument about technological unemployment in Africa has to begin by conceding something. On current evidence, the region is not experiencing large-scale automation. The World Bank’s October 2026 Africa Economic Update puts near-term automation risk in Sub-Saharan Africa at 2.6 per cent of jobs, against 14.2 per cent in high-income economies. By that measure, the region is among the least exposed on earth.
The structural reasons are well understood. The Bank’s earlier Future of Work in Africa found that the region’s much smaller manufacturing base means automation is unlikely to displace many workers in the coming years, and that informal employment accounts for almost 90 per cent of total employment. Informal firms do not buy industrial robots. Where labour is abundant and cheap and capital is scarce and expensive, the business case for automating is simply weak.
The empirical record points the same way. When fast internet reached the continent through submarine cables, Hjort and Poulsen found that employment rates rose rather than fell, including among less-educated workers.
It is worth being precise about why these figures do not contradict the headline estimates that put up to 85 per cent of jobs in some SSA economies at risk. Those estimates measure technical susceptibility, meaning how much of a job could in principle be automated. The World Bank’s 2.6 per cent measures something different, namely how much is likely to be automated in the near term given what firms can actually afford and deploy. Exposure describes what could happen. Adoption determines what does. This is why the paper treats occupation-based risk estimates as indicators of exposure rather than forecasts of job loss, and reconciles them against independent sources rather than reporting them at face value.
Why low exposure is not the same as safety
Here is the uncomfortable part. Every reason automation has not arrived is itself a symptom of underdevelopment. The region is insulated because its manufacturing base is small, its wages are low, its firms are informal, and its electricity and connectivity are unreliable. That is not resilience. It is distance from the productivity frontier. As ODI argues, limited exposure offers little reassurance if workers remain confined to poorly paid informal work while routes into higher productivity formal employment narrow.
The same World Bank update estimates the region’s augmentation potential, meaning the share of jobs that AI could make more productive rather than redundant, at 15.2 per cent, and ties that figure directly to connectivity. The opportunity is roughly six times larger than the near-term risk, and it is gated by infrastructure the region does not yet have. Low exposure today buys time. It does not buy an outcome.
Where the exposure actually sits
When adoption does come, it will not arrive evenly. Automation exposure in the region is concentrated in agriculture, where roughly 85 per cent of jobs are technically exposed, and in manufacturing, at around 70 per cent. These are not marginal sectors. Agriculture alone employs about 60 per cent of the workforce.
This is the vulnerability paradox at the heart of the transition: the sectors that absorb the most people are the sectors most exposed to being automated. In advanced economies, displaced workers can often move sideways into other industries. Here, absorptive capacity is thin, social protection is thinner, and a young population keeps arriving at the door.
The historical escape route has also narrowed. Rodrik has shown that developing economies now reach peak manufacturing employment at far lower income levels than earlier industrialisers did, a pattern of premature deindustrialisation that removes the labour-absorbing escalator on which previous transitions depended. The region is being asked to manage a technological transition without the ladder that made earlier ones survivable.
The burden is unevenly spread within the region too. Southern Africa combines the highest youth unemployment, at 47.5 per cent, with the highest NEET rate, at 34 per cent. Western Africa looks healthier on paper, with low measured unemployment, but that figure conceals widespread informal and insecure work alongside weak preparation for automation-resilient roles.
Resources are not the same thing as freedoms
This is where the capability approach, developed by Amartya Sen and Martha Nussbaum, earns its place. Its central insight is that a resource is not an opportunity. A laptop, a degree or an online course is a resource. It becomes a capability only when the holder can access electricity and connectivity, receives effective teaching, can exercise agency, can overcome discrimination, and meets real demand for their labour.
Those enabling conditions, which we call conversion factors, are exactly what is missing. Internet use in the region stands at 42.5 per cent against a world average of 68 per cent. Within Africa, urban use reaches 57 per cent while rural use sits at 23 per cent. Men are at 45 per cent and women at 31 per cent. A qualification earned under those conditions does not convert into the same thing.
Seen this way, the region is not facing a skills shortage so much as a capability deprivation. The distinction matters because it changes what counts as a solution. Train more people and the enrolment metric improves. Leave the conversion factors untouched, and their prospects do not.
Good strategies, weak systems
The second half of the problem is institutional. Across Kenya, Nigeria, Rwanda, South Africa and Ghana, we found national digital and education strategies that are, on the whole, well designed. The failure is rarely in the drafting. It is in the wiring.
Universities, firms, governments and international organisations are all present, but loosely coupled, operating in parallel rather than in concert. Knowledge generated in higher education seldom diffuses into the productive economy, a pattern consistent with the wider evidence on AI and workforce development. Donor priorities can crowd out locally determined ones. And implementation is persistently constrained by financing, administrative capacity, and a shortage of the disaggregated labour-market data that adaptive policy depends on. Where governments cannot see displacement happening, they cannot respond to it.
Rwanda and Ghana are partial exceptions, and instructively so. Both build monitoring and iteration into the policy cycle rather than treating a strategy as a finished document. Even there, urban and rural disparities persist.
The window, and what happens if it closes
Put the two halves together, and the stakes become clear. Adoption is currently slow, which means there is time. But adoption curves are rarely linear, and AI differs from industrial robotics in a way that matters here: it requires far less heavy capital, which weakens the very cost argument that has so far shielded the region.
If adoption accelerates before capability is built, the region would face displacement concentrated in its largest employers, with thin social protection, no manufacturing escalator to absorb the displaced, and more than 22 million new labour-market entrants every year. That is the catastrophic scenario, and nothing in the present trajectory rules it out. The modest exposure figures describe where the region is today, not where it is heading.
The implications split by audience. For universities, curricular relevance, faculty development, applied learning and genuine industry partnerships matter more than further growth in student numbers. For governments, educational reform must be sequenced alongside investment in connectivity, electricity, and social protection, because the education sector cannot deliver outcomes that depend on conditions outside it. For employers and development partners, the highest-value contribution is co-designing curricula, providing placements, and strengthening local agendas rather than substituting for them. And investment in national statistical capacity is not an administrative detail. It is a precondition for governing this transition at all.
The question facing the region is not whether automation will arrive. It is whether higher education and the institutions around it can build the capabilities that let people benefit when it does. On present evidence, the gap is persistent. It is not, however, inevitable.
Read the paper: Ekuma, K., & Uba, C. D. (2026). The Persistent Capability Gap: Higher Education, Innovation Systems and Technological Unemployment in Sub-Saharan Africa. Cogent Education. https://doi.org/10.1080/2331186X.2026.2745299
Dr Kelechi Ekuma is based at the Global Development Institute, The University of Manchester. Dr Chijioke Dike Uba is in the Department of Strategy and International Business, Faculty of Business and Law, University of Northampton.
A note on sources
Unless stated otherwise, the figures in this piece are drawn from the underlying paper, Ekuma and Uba (2026), where each is fully cited. The primary sources are: youth unemployment and NEET rates, ILO, Global Employment Trends for Youth 2024; internet use and the urban, rural and gender gaps, ITU, Measuring Digital Development: Facts and Figures 2024; advanced digital skills and STEM competencies, UNDP, Human Development Report 2023/24; and automation exposure, sectoral employment shares and labour-market entrants, World Bank, Digital Africa: Technological Transformation for Jobs (2023).
Three figures are not drawn from the paper and are linked in the text where they appear: near-term automation risk of 2.6 per cent and augmentation potential of 15.2 per cent, from the World Bank’s Africa Economic Update, October 2026; and informal employment at almost 90 per cent of total employment, from the World Bank’s The Future of Work in Africa (2019).
Photo by Christian Agbede on Unsplash
Note: This article gives the views of the author/academic featured and does not necessarily represent the views of the Global Development Institute as a whole.
Please feel free to use this post under the following Creative Commons license: Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0). Full information is available here.