Bank of Korea AI Productivity Gap
3 steps1 pieces6 Sources
The Bank of Korea (BOK) has identified a significant "productivity disconnect" in South Korea's adoption of generative AI. While workplace usage surged to 66.5% in 2026—more than double the average of six European countries and far exceeding the US at 43.0%—the time saved by workers is not translating into measurable gains in output or quality. A BOK survey of 16,342 employed individuals aged 18–64 revealed that although AI users saved an average of 9.2% of their working time (about 3.7 hours per week), the correlation between these savings and increased work volume was only 0.08, and with improved work quality, just 0.09.
This gap highlights that while AI has reached an "efficiency stage," it has not yet progressed to a "productivity stage." The central bank attributes this to insufficient organizational restructuring and workflow redesign. Notably, productivity gains were observed primarily among self-employed professionals and heavy users with greater autonomy, suggesting that incentive structures matter more than mere adoption rates. With annual worker spending on AI subscriptions estimated at 2.8 trillion won against a potential economic value of 61.1 trillion won in saved time, the BOK urges policymakers to shift focus from expanding access to enhancing the depth of integration into core business tasks.
How it got here
June 8, 2026: The Bank of Korea releases Issue Note No. 2026-12, 'Does AI Adoption Improve Productivity?', analyzing data from May–June 2025. It finds a 3.8% reduction in work time but zero correlation with output growth, coining the term 'AI productivity disconnect.'
bok.or.kr ↗November 17, 2026: The BOK Employment Research Team publishes a new report, 'AI Diffusion and Productivity Effects,' based on a survey conducted August 14–September 23, 2026. It shows generative AI usage rose to 66.5%, with time savings increasing to 9.2%, but correlations with productivity remain weak (0.08 for volume, 0.09 for quality).