Key findings
- Unemployment tells you almost nothing about AI exposure. Across 186 countries the correlation is 0.04. For youth unemployment it is 0.19, still weak (World Bank WDI, ILO ILOSTAT and WorldJobsData).
- Nine countries carry both burdens. 9 countries have unemployment of at least 10% and AI exposure of at least 4.5: South Africa, Djibouti, Namibia, Saint Vincent and the Grenadines, Montenegro and 4 more.
- South Africa is the extreme case. Unemployment is 32.4%, youth unemployment 59.9%, and AI exposure 4.84.
- Most high-unemployment countries are less exposed. 29 of the 38 countries with unemployment of 10% or more score under 4.5. Their problem is too few jobs, not AI.
- Young people face higher rates almost everywhere. In 178 of 181 countries the youth rate is higher than the overall rate. The median youth rate is 12.7%, against 5.1% overall.
Two numbers that are easy to mix up
When people talk about "AI and jobs," they often assume that countries with weak job markets are the ones most at risk. These are really two different questions. The unemployment rate is the share of people in the labour force who want a job and cannot find one. AI exposure is our own score from 0 to 10 for how much of a country's work could be done or helped by AI. One describes today's job market. The other describes the type of work in a country.
So it is worth testing whether they go together. If countries with high unemployment were also more exposed, AI would be hitting already weak job markets. If they do not go together, the two problems are separate and need different answers.
The test: no pattern at all
The dots spread across the whole chart. Countries with 2% unemployment can score 3 or 5. Countries with 10% unemployment can too. The correlation is 0.04, which is as close to nothing as real data gets. For youth unemployment (ages 15 to 24) the link is 0.19, still weak. Overall and youth unemployment themselves move together very closely (0.93), so the youth numbers add little new information on this question.
Who has the highest unemployment?
Four of the top ten are in southern Africa. Eswatini (34.2%), South Africa (32.4%), Botswana (24.5%) and Namibia (19.3%) all have more than one in five workers unemployed. Djibouti and Palestine are also in the top five. Of these, South Africa, Djibouti and Namibia have AI exposure above 4.5.
Youth unemployment is far higher. In Djibouti 76.8% of young people in the labour force have no job. In South Africa it is 59.9%. Young workers usually start in junior roles, and routine tasks in those roles are where AI assistants can take on work.
Where the two problems overlap
| Exposure 4.5 or more | Exposure under 4.5 | |
|---|---|---|
| Unemployment 10% or more | 9 | 29 |
| Unemployment under 10% | 47 | 101 |
Splitting countries at unemployment 10% and exposure 4.5 gives four groups. The group with both high unemployment and high exposure is small, but it is the group to watch:
| Country | Unemployment | Youth unemployment | AI exposure | Workers covered |
|---|---|---|---|---|
| South Africa | 32.4% | 59.9% | 4.84 | 12.7M |
| Djibouti | 26.0% | 76.8% | 4.51 | 0.1M |
| Namibia | 19.3% | 38.0% | 4.62 | 0.5M |
| Saint Vincent and the Grenadines | 18.0% | 42.0% | 4.93 | 0.0M |
| Montenegro | 13.6% | 24.4% | 4.69 | 0.2M |
| French Polynesia | 11.7% | 36.6% | 5.09 | 0.1M |
| US Virgin Islands | 11.4% | 24.2% | 4.99 | 0.0M |
| New Caledonia | 11.2% | 33.1% | 5.25 | 0.0M |
| Spain | 10.4% | 24.8% | 4.72 | 22.2M |
These are not alike. South Africa and Namibia are middle-income countries with large unemployed populations and sizeable office sectors. Spain (10.4% unemployment, 24.8% for the young) is the one large European economy here. New Caledonia, French Polynesia, Montenegro and the US Virgin Islands are small economies, so their averages are based on small workforces. In all of these, junior workers who would normally enter office jobs face both a weak market and tools that can do some of their entry-level tasks.
The large economies
| Country | Unemployment | Youth unemployment | AI exposure |
|---|---|---|---|
| South Africa | 32.4% | 59.9% | 4.84 |
| Spain | 10.4% | 24.8% | 4.72 |
| Greece | 8.5% | 21.4% | 4.88 |
| Italy | 6.4% | 20.6% | 4.82 |
| France | 7.5% | 18.9% | 5.06 |
| Egypt | 6.8% | 18.3% | 3.87 |
| Turkiye | 8.5% | 16.0% | 4.11 |
| India | 4.2% | 16.0% | 3.26 |
| China | 4.6% | 15.8% | 4.48 |
| United Kingdom | 4.8% | 14.7% | 5.08 |
| Brazil | 6.0% | 14.0% | 4.13 |
| United States | 4.2% | 9.3% | 5.06 |
| Germany | 3.7% | 6.9% | 5.30 |
| South Korea | 2.7% | 6.7% | 4.85 |
| Nigeria | 3.1% | 5.3% | 3.31 |
| Japan | 2.5% | 3.9% | 4.92 |
Among the biggest economies, Spain, Greece and Italy combine youth unemployment above 20% with AI exposure of about 4.7 to 4.9. Japan (3.9%), South Korea (6.7%) and Germany (6.9%) have the lowest youth rates among the large economies here while being highly exposed (4.92, 4.85 and 5.30), which suggests that exposure on its own does not produce unemployment. India has youth unemployment of 16.0% but exposure of only 3.26.
What this means
The practical message is to treat unemployment and AI exposure as two separate risks. In most high-unemployment countries, the urgent problem is too few jobs, and a low AI score does not help someone who cannot find work. In the handful of countries with both, such as South Africa and Spain, the entry-level pathway into office work deserves special attention, because that is where young jobseekers and AI tools overlap.
For anyone in a high-exposure country, the unemployment rate today is not a prediction of AI disruption tomorrow. Germany and Japan show that very high exposure can coexist with some of the lowest unemployment in the world.
Limits to keep in mind
- Different measures. Unemployment is a rate of people; exposure is a score of tasks. They describe different things by design.
- Years. Unemployment is mostly 2025 (181 of 186 countries), with the rest from 2021 to 2023. Youth unemployment is the latest ILOSTAT year, here 2025.
- Small labour markets. Small countries can swing widely from year to year, so treat very high rates with care.
- No cause shown. Nothing here says AI has caused any unemployment so far.
See your country's job market and AI scores
Open a country page for unemployment, youth unemployment, informality and occupation-level AI exposure.
Open the explorer →wb_economy.unemployment_pct), 181 of 186 countries at 2025. Youth unemployment (ages 15 to 24) is the ILO ILOSTAT rate UNE_2EAP_SEX_AGE_RT, latest year, stored in meta.json, for 181 countries. AI exposure is each country's weighted average exposure (stats.weighted_avg_ai_exposure). Correlation is Pearson's r. The four-group split uses thresholds of 10% unemployment and 4.5 exposure, which we chose and which are not official categories. Calculations are ours.
AI exposure scores are model-estimated: an AI language model (Claude Opus 4.7) scored each ISCO-08 occupation group on 28 May 2026, guided by published research. They are informed judgements, not measurements, and not forecasts of job losses. Read the full methodology and how the scores were checked.
Frequently asked questions
Does high unemployment mean high AI exposure?
Which countries combine high unemployment and high AI exposure?
What is youth unemployment?
Is AI the cause of high unemployment?
Get new country analyses, data updates, and AI labour market insights. No spam - one email when something worth reading drops.
Sources
- World Bank (2026) - World Development Indicators, unemployment, total (% of labour force), modeled ILO estimate (CC BY 4.0)
- ILO (2026) - ILOSTAT, youth unemployment rate (ages 15-24), latest available year (CC BY 4.0)
- WorldJobsData (2026) - Occupation-level AI exposure scores, methodology
Spotted an error or know a better source? Report it and we will check it against the source data.