At a glance
- Source
- Georgieff and Hyee (2022). Artificial Intelligence and Employment: New Cross-Country Evidence. Frontiers in Artificial Intelligence, 5, 832736. doi:10.3389/frai.2022.832736
- How it was done
- The authors took an existing score of how exposed each occupation is to AI, extended it to 23 countries, and compared it with changes in employment and hours. Data analysis. It builds a measure from existing data, such as job and task records.
- Who took part
- Occupations in 23 OECD countries.
- Where
- 23 member countries of the OECD, a group of mostly high-income economies.
“Overall, there appears to be no clear relationship between AI exposure and employment growth.”
What it found
- Overall, exposure to AI did not predict whether employment in an occupation grew or shrank.
- In occupations where people use computers a lot, higher exposure went with higher employment growth.
- In occupations with little computer use, there were signs that higher exposure went with fewer hours worked.
The authors suggest one explanation: workers with good digital skills can use AI to move to higher-value tasks, while workers without those skills cannot.
When the result holds
The data cover the years before today’s AI chat tools. The study shows what went together across countries. It does not prove cause.
Later US data on young workers in exposed jobs point the other way. See fewer young people hired into AI-exposed jobs.
Limits
- It compares occupations, not companies.
Practices that use this study
This study supports: