Does AI Adoption Improve Productivity? Effects over the First Three Years
A Bank of Korea study finds that AI adoption boosts firms' productivity by an average of 2.1% over the first three years, with the effect being more pronounced in services than in manufacturing. However, the benefits take time to materialize, showing no significant impact in the first year, and are concentrated among firms with higher AI readiness and data infrastructure.
Background
- This is a research note from the **Bank of Korea** (BOK), South Korea's central bank (analogous to the Federal Reserve in the US or the ECB in Europe). BOK regularly publishes economic analysis papers to inform policy.
- The question "Does AI adoption improve productivity?" is central to a major global debate: AI has been hyped for years, but economists are still trying to measure whether it actually boosts output per worker in a measurable way — something earlier technologies (like the internet) eventually did.
- The note specifies effects "over the first three years", meaning it looks at short-run impacts. This is important because previous general-purpose technologies (like electricity or computers) often showed a "productivity paradox" — initial adoption costs and organizational disruption meant productivity gains took much longer to appear than expected.
- The intended audience is policymakers and economists; the key takeaway is whether South Korea — a highly digitized, manufacturing-heavy economy — is seeing concrete returns on AI investment, or whether the benefits are still delayed or unevenly distributed across sectors.