• The ILO confirms that AI is transforming employment at different rates across regions.
  • In developing economies, exposure to automation is lower, but disruptions can occur before productivity gains materialize.
  • Advanced economies face greater exposure: some observed effects come through reduced hiring, while data on AI-related layoffs remains limited.
  • In Africa, fewer than 0.6% of job postings mention AI-related skills, compared to fewer than 2% in the United States and Europe according to the World Bank.

The essentials

  • The ILO notes lower aggregate exposure to automation in developing economies, with benefits constrained by digital gaps, but emphasizes that rapid AI diffusion requires swift action.
  • In Africa, fewer than 0.6% of job postings mention AI-related skills, compared to 1 to 2% in the United States and Europe (World Bank).
  • The major risk is not immediate technological unemployment, but value capture by foreign platforms and infrastructure if structural choices are delayed.
  • Several African countries—Kenya, Rwanda, Senegal, Egypt—have launched national AI strategies, but the gap between policy documents and actual capacity remains wide.
  • The fundamental question is who on the continent will train workers, lay cables, write the rules, and capture the revenues of the forthcoming digital economy.

Delay is a fact; irreversibility is not yet

The 0.6% figure measures the share of online job postings observed in African countries studied. It is an indicator of formal labor market demand, not a technological unemployment rate or a measure of disengagement. It signals that African job postings still rarely mention AI-related skills, which does not allow conclusions about automation in the recruitment process itself.

In the United States, this rate is below 2% and has been rising since 2020. In Europe, it sits below 2%, with peaks in the financial and logistics sectors. The gap is real. But it is not fixed.

The ILO, in its November 2026 report, stresses a point that automation discussions often neglect: AI diffusion within organizations is slow. Documented productivity gains in advanced economies concentrate in certain sectors and remain limited in light manufacturing, agriculture, or proximity services that employ the majority of African workers. AI diffusion is rapid compared to previous technologies, but deep adoption and the necessary capabilities remain unevenly distributed.

For Africa, this means something concrete: structural employment transformations are not imminent in the sense they would be if diffusion had been as fast as public discourse suggested. The continent has a window. Not a guarantee.

The window for decision: scope and limits

A window for decision has value only if decisions are actually made. The African situation is more nuanced in this regard.

Several governments grasped the stakes relatively early. Egypt adopted a national AI strategy in 2019, since revised. Kenya published a digital roadmap that includes AI as a pillar of public service development. Rwanda is positioning Kigali as a regional technology hub, with investments in connectivity and training. Senegal launched a digital transformation plan in 2023 that includes skills programs for young people.

These initiatives exist. They are real. But the World Bank signals a persistent gap between national strategies and actual capacity: lack of qualified trainers, insufficient electrical infrastructure to operate local data centers, gaps in data protection, cybersecurity, regulation, and enforcement capacity in many developing countries.

Governments have grasped the stakes but face capacity deficits: lack of qualified trainers, insufficient electrical infrastructure, absence of operational regulatory frameworks. An AI strategy must be accompanied by investments and implementation capacity in electrical infrastructure, engineer training, data, computing power, and institutions to produce large-scale effects. The ILO recommends investments in infrastructure and skills, worker protections, support for local businesses, and cooperation on data, computing, competition, and governance.

The informal sector: neither sheltered nor out of play

A common mistake is thinking that Africa’s informal sector, which employs between 70 and 90% of the workforce depending on the country, is insulated from AI effects because it is barely digitized. This intuition is partly right and partly misleading.

It is right because a street vendor in Dakar or an artisan in Kampala is not directly exposed to cognitive automation of their tasks. AI does not enter their working life through an employer’s decision, because they have no employer.

It is misleading because AI transforms the markets on which these workers operate. E-commerce platforms reshape distribution circuits. Mobile payment systems, already dominant in East Africa, integrate credit algorithms that decide who can borrow for their business. Transportation and delivery applications restructure task-based work according to optimization logic that workers do not control, a phenomenon already documented in the context of algorithmic work in Europe and beginning to roll out in major African cities.

African informal work is exposed differently by digital transformations, even with low levels of direct digitization. It is exposed differently: market reorganization and margin capture by digital intermediaries create particular structural risks for informal workers. This indirect AI effect escapes standard measures, such as the share of job postings mentioning AI skills.

Data, cables, and rules: whoever controls infrastructure decides

The question of value capture is probably the most structuring issue for Africa over the next fifteen years.

The AI economy rests on three pillars: data, computing infrastructure, and the rules that govern their use. On all three dimensions, the African continent starts in a position of dependence.

Data generated on the continent—mobile financial transactions, agricultural data, health data, urban transport flows—are processed on servers whose location depends heavily on limited local capacity and recourse to international cloud services. Some cables and consortiums rely on foreign actors or international financing, but African cable ownership varies widely and includes majority African holdings. Africa remains highly dependent on systems designed elsewhere, but it also produces local AI applications, services, and innovations, such as those developed for isiZulu and Sesotho.

This infrastructural dependence is not inevitable. Projects for African-financed submarine cables are emerging. The African Union adopted a convention on cybersecurity and personal data protection, the Malabo Convention, though ratification remains incomplete. Local players like Andela, Flutterwave, or Zipline are building economic models anchored to the continent.

But the window for decision on rules is short. Technical standards and AI regulatory frameworks are being defined at the European Union level with the AI Act, in the United States, in China. Africa that does not enter these negotiations with its own position will find itself applying rules designed for other contexts. This is what happened with intellectual property regimes from trade agreements in the 1990s, whose effects on African pharmaceutical and cultural industries have been documented.

Scenarios for 2030 based on choices made now

Several trajectories are possible by 2030-2035, depending on decisions made in the next five years. None is certain; both are conditional.

In the first, African countries that have invested in skills, stabilized their electrical infrastructure, and adopted data regulations could capture an increased share of the value generated by digital work on their territory. Regional hubs—Nairobi, Lagos, Accra, Kigali, Tunis—become nodes in an African AI economy capable of producing models adapted to local languages and contexts. Digital task work, governed by rules protecting worker earnings, becomes a vector for income growth rather than a new form of precarity. Articles on the geographic concentration of AI jobs in Europe and the United States show that such concentration is possible wherever basic conditions are met.

In another trajectory, the window closes without substantial investments and AI skill adoption advances slowly. Platforms may hold market power and regulatory gaps may expose workers to risks, depending on sectors and countries. Africa remains dependent on imported systems, and the skills gap deepens.

These trajectories carry different probabilities based on investment and regulation choices. The current trend, marked by weak public investment in technical training, slow ratification of data protection frameworks, and dependence on foreign infrastructure, creates risks of tilting toward the second trajectory. But several signals point elsewhere: the explosion in the number of African developers trained by programs like ALX Africa, the rise of national data regulators, the first agricultural data-sharing agreements at the regional scale.

The actors moving things forward now

The ILO, in its November 2026 report, does not merely diagnose. It documents ongoing experiments.

In Kenya, the TVET program, technical and vocational training, offers a distinct AI curriculum. The objective is to train workers capable of using AI tools in their trades, not necessarily designing them. This is a pragmatic approach: close the skills gap through use before investing in basic research.

In Egypt, the Ministry of Communications launched partnerships with technology companies to create recognized certifications in machine learning and data processing. In Rwanda, DOT Rwanda is implementing with the Mastercard Foundation and local organizations a digital training program targeting more than 10,000 young people by 2028, with an emphasis on applied AI skills.

African private actors are also building solutions. Lelapa AI, based in South Africa, develops linguistic tools for African languages such as isiZulu and Sesotho to fill the blind spot of major global models on the continent’s linguistic contexts. Masakhane, a collective of African researchers, works on datasets and AI tools built for and by speakers of African languages.

These initiatives remain modest relative to the scale of the challenge. They exist, they are African, and they show that the decision window is open beyond the government level alone.

The collective decision still to be made

AI will diffuse in Africa. The current slowness documented by the ILO does not mean the continent can wait: structural employment transformations are not immediate, and this window has value only if foundations are laid during this period.

These foundations are known: reliable electricity for computing infrastructure, large-scale technical training, operational data protection frameworks, digital work rules that organize revenue sharing, African participation in international negotiations on AI standards.

None of these conditions is remarkable. They are the same conditions that countries succeeding in their technological transitions have met, at different moments in history. The question is whether African decision-makers—governments, regulators, universities, businesses—will treat this window as a structural opportunity or as a problem to manage when it becomes urgent.

World Bank data and the ILO report give the same information to both types of decision-makers. What will differentiate them in ten years is what they will have done with it.

Sources

  1. International Labour Organization, AI and Developing Economies: Promoting Decent Work in a Time of Rapid Technological Change (November 2026)
  2. World Bank, Report on AI-Related Skills in Developing Economies
  3. International Labour Organization, Malabo Convention on Cybersecurity and Personal Data Protection (African Union)
  4. ALX Africa, digital skills training program
  5. Lelapa AI, language models for African languages (Cape Town, South Africa)
  6. Masakhane, AI research collective for African languages