Enterprise & Work

Bank of Korea: 66.5% of workers use AI, productivity gains lag

Bank of Korea surveys of 16,000-plus people show AI adoption at 66.5 percent and weekly use at eight hours, yet no clear correlation between time saved and gains in work volume or quality.

AI usage spreads rapidly among S. Korean workers, but productivity gains remain limited: report
AI usage spreads rapidly among S. Korean workers, but productivity gains remain limited: reportAI-generated
By Rebecca Stone5 min read

Why it matters

  • 66.5 percent of surveyed South Korean employees used generative AI in 2026, up from 51.8 percent in 2025.
  • Weekly AI usage rose to 8 hours from 5.9 hours; reported time savings on work tasks reached 9.2 percent, up from 3.8 percent.
  • 28.4 percent of AI users paid for services in 2026, averaging 49,000 won ($37) per month.
  • South Korea's 66.5 percent adoption rate exceeded the United States (43 percent) and Germany (31.5 percent).
  • The Bank of Korea found no clear correlation between time saved and increases in work volume or quality.

Two-thirds of South Korean employees now use generative AI tools, yet the time they save produces no measurable rise in work volume or quality, the Bank of Korea reported Tuesday.

The central bank's report, based on surveys of more than 16,000 people conducted in 2025 and 2026, found 66.5 percent of surveyed employees used at least one generative AI tool this year — including ChatGPT, Gemini and Claude — up from 51.8 percent a year earlier. Weekly usage climbed to eight hours from 5.9 hours over the same period.

The stakes are straightforward. South Korea has positioned itself as one of the fastest adopters of workplace AI among advanced economies, and the Bank of Korea's data now offers one of the larger employee-level evidence bases on whether that adoption pays off. So far, the report concludes, it does not show up in output.

What did the surveys measure?

The Bank of Korea surveyed more than 16,000 people across 2025 and 2026, tracking four main indicators:

  • Adoption rate: 66.5 percent of employees used generative AI in 2026, up from 51.8 percent in 2025.
  • Usage intensity: average weekly AI use rose to eight hours from 5.9 hours.
  • Time savings: respondents said AI cut time spent on work tasks by 9.2 percent in 2026, compared with 3.8 percent in 2025.
  • Paid subscriptions: 28.4 percent of AI users paid for AI services in 2026, spending an average of 49,000 won ($37) per month.

The 9.2 percent time-saving figure more than doubled year over year. The report also found stronger time-saving effects among higher-educated workers, a gap that suggests the benefits of the technology are unevenly distributed across the workforce even as adoption spreads.

How does South Korea compare internationally?

The 2026 adoption rate of 66.5 percent exceeded corresponding rates the Bank of Korea cited from separate research for the United States, at 43 percent, and Germany, at 31.5 percent.

South Korea's rate is more than one and a half times the US figure and more than double Germany's. That comparison matters for policy: if the world's fastest-adopting workforce cannot convert AI usage into measured productivity, the finding complicates the assumption — common among governments and vendors — that diffusion of the tools by itself lifts economic output.

Why haven't time savings become productivity gains?

The report's central finding is the disconnect between the hours reclaimed and the work produced. Respondents reported using generative AI for a growing share of their week, and they credited it with cutting task time by 9.2 percent. But the Bank of Korea found no clear payoff in either quantity or quality of work.

"Although many respondents indicated that they used the saved time to focus on core tasks or handle additional work, no clear correlation was found between the rate of time savings and the rates of increase in work volume or quality improvement," the report's findings showed.

In other words, workers believe AI saves them time, and many say they redirect that time toward core tasks or extra work. But at the level the Bank of Korea measured, the saved minutes do not translate into more output or better output. The report does not claim the gains are zero; it states that no clear correlation emerged between time saved and increases in work volume or quality.

The pattern echoes a broader puzzle in productivity statistics: individual tools can feel time-saving to their users while aggregate measures of output remain flat. The Bank of Korea's survey, with its two-year window and 16,000-plus respondents, adds a data point from the economy adopting AI fastest.

Who is spending money on AI?

Paying for AI remains a minority behavior, but a growing and quantifiable one. Among AI users surveyed in 2026, 28.4 percent subscribed to paid AI services, at an average cost of 49,000 won — about $37 — per month.

That spending figure gives the report a concrete market dimension. If adoption continues to rise from 66.5 percent, and if the paid-subscription share among users grows beyond 28.4 percent, monthly consumer spending on AI tools in South Korea scales quickly across a workforce of millions. Whether that spending earns a return is precisely the question the report leaves open.

What are the report's limits?

The findings rest on self-reported data. Respondents estimated their own AI usage hours, their own time savings, and their own assessments of work volume and quality. The Bank of Korea then tested whether reported time savings correlated with reported increases in output and quality, and found no clear relationship.

The report also covers two years, 2025 and 2026, a short window for a technology whose workplace effects may take longer to appear. The education gap in time savings — stronger effects for higher-educated workers — suggests that complementarities between AI and worker skills may still be forming.

What happens next?

The Bank of Korea has now put a number on both sides of the AI productivity debate: 66.5 percent adoption and eight hours of weekly use on one side, and no clear correlation between time saved and work volume or quality on the other. If next year's survey shows the time-saving figure rising further without a matching rise in measured output, the burden of proof shifts to those who claim workplace AI is already lifting productivity rather than merely redistributing workers' hours.

Source: feed.koreatimes.co.kr

Share this article:

More from Rebecca Stone

Rebecca Stone

Show full bio

Correspondent covering companies, deals and strategy in artificial intelligence and machine learning at AI In Context.

728 articles

Related articles

  1. 67.8% of Workers Now Use AI Weekly — But 56.4% Get No Time to Learn It
  2. KDI raises South Korea’s 2027 growth forecast to 3.9%
  3. AI Users Work Up to 90 Minutes Longer Than Colleagues, ActivTrak Study Finds
  4. OpenAI Study: Researchers Say ChatGPT Cuts Their Work Time Nearly in Half
  5. 95 Percent of Organizations See No Measurable AI Gains, University Leaders Warn

« Previous articleNext article »