1 job in 4
Global share of jobs exposed to generative AI · estimate
ILO, 2025 ↗
34% versus 11%
Share of jobs exposed to generative AI: 34% in high-income countries, 11% in low-income countries.
Share of jobs exposed, by country income level · estimate
ILO, 2025 ↗
What AI touches in an occupation
Each square is one of the occupation’s tasks; hover over one for details.
- Mostly delegated to AI
- Mostly done with AI
- Not in the data
- Computer programmers11 / 17 tasksComputer programmers: 11 of 17 tasks appear (10 mostly delegated to AI, 1 mostly done with AI).
- Tutors10 / 19 tasksTutors: 10 of 19 tasks appear (3 mostly delegated to AI, 7 mostly done with AI).
- Accountants and auditors11 / 29 tasksAccountants and auditors: 11 of 29 tasks appear (7 mostly delegated to AI, 4 mostly done with AI).
- Secretaries and administrative assistants10 / 31 tasksSecretaries and administrative assistants: 10 of 31 tasks appear (8 mostly delegated to AI, 2 mostly done with AI).
- Customer service representatives4 / 13 tasksCustomer service representatives: 4 of 13 tasks appear (2 mostly delegated to AI, 2 mostly done with AI).
- Farmers3 / 30 tasksFarmers: 3 of 30 tasks appear (2 mostly delegated to AI, 1 mostly done with AI).
- Nurses2 / 27 tasksNurses: 2 of 27 tasks appear (2 mostly delegated to AI).
- Carpenters1 / 29 tasksCarpenters: 1 of 29 tasks appears (1 mostly delegated to AI).
O*NET database tasks found in work-related conversations with an AI assistant (Claude), worldwide, May 2026 · observed data
Anthropic Economic Index, 2026 ↗
What determines the capacity to act
- Access to electricity
- Networks
- Computing power
- Skills
- Representative data
- Tools adapted to local languages
↓ Capacity to act
AI is a general-purpose technology, with effects comparable to those of the great industrial revolutions. It is transforming research, design, creation and innovation; combined with robotics, it also affects production, logistics and services. Source: OECD, 2026 ↗
The ILO estimates that about one job in four worldwide is exposed to generative AI. Only 3.3% of jobs, however, fall into the highest exposure category: most occupations will be transformed rather than eliminated. Clerical jobs are the most affected, and women more than men. Exposure reaches 34% of jobs in high-income countries, against 11% in low-income countries. For Africa, the challenge is therefore twofold: preparing for this transformation and gaining access to the benefits it can bring. Beyond occupations and skills, this shift raises questions about the role of work, time, learning and the value of each contribution. It may alter organisations and social relations, as well as our relationship to autonomy, creativity, recognition and psychological well-being. How can its gains be shared without widening inequalities? Source: ILO, 2025 ↗
Actual usage shows how this divide takes shape, task by task. Eleven of a computer programmer’s 17 tasks already appear in conversations with an AI assistant, most often delegated to it; for a tutor, 10 of 19, mostly done with it. For a farmer, a nurse or a carpenter, only one to three tasks out of around thirty appear. In the United States, about 30% of workers do not yet have a single task concerned. AI is first transforming office, writing and analytical work; occupations based on manual work and human contact remain, for now, far less affected. Source: Anthropic Economic Index, 2026 ↗ Source: Anthropic, March 2026 ↗
Its effects will also depend on collective choices. Access to electricity, networks, computing power, skills, representative data and tools adapted to local languages all determine the capacity to act. Who designs these systems and sets their rules? Who can challenge their decisions and benefit from the value created? For Africa, technological sovereignty is one dimension of this revolution, alongside its economic and human consequences. Source: African Union, 2024 ↗