Kim Yeon-jae
Kim Yeon-jae is a journalist at AJU Press (AJP's English platform),
covering macroeconomics, international finance, and geopolitics.
He closely tracks central bank monetary policies, global energy supply chains,
and the Korean defense industry.
Copyright ⓒ Aju Press All rights reserved.
Related News
-
More workers use AI, but productivity remains low: BOK study
Edited and compiled by AJP. Graphics AI assisted. SEOUL, November 17 (AJP) — About two in three South Korean workers now use generative artificial intelligence (AI) at work, but the time saved has yet to translate consistently into higher output, a recent study reveals on Tuesday. Workplace AI adoption rose to 66.5 percent this year from 51.8 percent in 2025, the highest rate among countries compared in a Bank of Korea (BOK) study conducted by researchers Seo Dong-hyun, Oh Sam-il and Yoon Jong-won. The figure was nearly double the 33.7 percent average for the U.S. and six European countries. The average share of working time saved through AI more than doubled to 9.2 percent from 3.8 percent a year earlier. That translates into about 3.7 hours saved in a 40-hour workweek, compared with 1.5 hours last year. Yet the correlation between time saved and increased work output stood at just 0.08, up from zero in 2025, suggesting that individual efficiency gains have not consistently translated into greater productivity. The researchers cautioned that the weak correlation did not mean AI had failed to increase output. Rather, workers who saved more time were not necessarily completing proportionately more tasks. Some 33.8 percent of respondents said they primarily used the time saved to concentrate on core duties, while 29.3 percent reported taking on additional work. Another 13.2 percent cited learning and professional development. The researchers noted that faster task completion may not increase output if the amount of work assigned or demand for additional services remains unchanged. They pointed to the need for changes in workflows, task allocation and workplace organization to translate individual efficiency gains into higher overall output. AI adoption also remained uneven across occupations. Usage reached 86.0 percent among professionals and 81.3 percent among office workers, compared with 39.6 percent among elementary workers. Even in occupations with high adoption rates, actual AI use in core duties fell substantially short of what researchers assessed as technically possible. Paid access was associated with stronger efficiency gains. Workers who personally paid for AI services reported average time savings of 11.4 percent, compared with 8.3 percent among those who did not pay out of pocket. Employees receiving company support reported savings of 11.6 percent. The researchers cautioned that the differences could not be attributed solely to paid access, as workers with greater demand for AI tools may be more likely to subscribe. The study also found limited evidence that longer experience with AI or formal training had significantly improved time savings over the past year, suggesting that advances in AI technology itself may have played a substantial role. Workers who actively experimented with AI tools, checked results and integrated them into their routines, however, showed stronger improvements. The researchers estimated that AI could potentially increase aggregate output by 2.6 percent if all time saved were redirected to productive activities, up from an estimated 1.0 percent last year. That figure represents a hypothetical gain under the study's assumptions, not an increase in actual gross domestic product. The relationship between time savings and improvements in work quality was also weak, with a correlation coefficient of 0.09. However, workers whose duties involved a larger share of core tasks reported relatively stronger quality gains. The findings suggest that some AI benefits may emerge through better results rather than faster task completion or greater work volumes. The study was based on a survey of 16,342 employed people aged 18 to 64 conducted between August and September, alongside panel data tracking 2,024 respondents from the previous year's survey. The researchers concluded that wider adoption alone would be insufficient to realize AI's full economic potential, highlighting the importance of improving workers' skills and adapting how companies organize and assign work.
-
South Korea's chip boom widens wage divide: KDI
Source: Korea Development Institute (KDI) report. SEOUL, November 16 (AJP) - South Korea's semiconductor productivity rose 8.5 percent annually from 2021 to 2025, while productivity in other industries fell 0.4 percent a year to deepen wage gap, the Korea Development Institute (KDI) said Monday. The report found that weak productivity outside the semiconductor sector was a major reason real wages, adjusted for inflation, have barely risen since 2021. In the first half of 2026, average real wages rose just 0.3 percent from a year earlier, while inflation-adjusted regular pay fell 0.3 percent. Workers in industries where average real wages declined accounted for 53.2 percent of the workforce. Source: Korea Development Institute (KDI) report. Graphics AI-assisted. Wage gains were concentrated in a few high-paying industries. Nominal wages rose 21.6 percent in electronics manufacturing, which includes semiconductors, and 12.3 percent in financial and insurance-related services, with bonuses and other special payments accounting for much of the increase. The broader wage stagnation has persisted for years. Average real wages across all workers rose just 0.2 percent from 2021 to 2025, compared with a 24 percent increase over the previous decade. The report linked the wage gap to a widening productivity divide. From 2011 to 2025, semiconductor productivity increased an average of 11.3 percent annually, compared with just 1 percent in non-semiconductor industries. Source: Korea Development Institute (KDI) report. Graphics AI-assisted. KDI measured productivity using company financial statements to assess how much value firms generate from labor and capital. The measure reflects technological improvements as well as changes in demand and selling prices. That divide was also reflected in regular pay. From 2021 to 2025, inflation-adjusted regular pay in electronics manufacturing grew about 9 percentage points faster than in other industries. KDI estimated that differences in productivity growth accounted for roughly one-third of that gap. Industries with 10 percentage points higher cumulative productivity growth from 2021 to 2025 saw inflation-adjusted regular pay rise an average of 0.9 percentage points more than in other industries, according to the report. The weakness extended even to more productive companies outside the chip industry. Among non-semiconductor firms, the top third by productivity recorded a cumulative increase of just 0.8 percent from 2021 to 2025. Productivity in the middle third fell 2.5 percent, while the bottom third recorded a 9.7 percent decline. The prolonged weakness has also left an increased share of non-semiconductor firms struggling financially. The share of marginal firms, measured by value added, in non-semiconductor industries nearly doubled from 3.2 percent in 2021 to 6.2 percent in 2025. KDI attributed much of the increase in marginal firms in industries with the lowest productivity growth to poor productivity rather than higher borrowing costs. Without broader productivity gains, KDI warned, wage increases could remain concentrated in a handful of industries. It called for existing industries to develop higher-value products and for barriers to entry and growth in emerging industries to be removed. The institute also urged policymakers to prioritize business restructuring and worker retraining over blanket financial support, such as loan extensions and interest subsidies, to help labor and capital move toward more productive businesses. Tags: South Korea wage gap, KDI productivity report, semiconductor industry, real wage stagnation, non-semiconductor productivity
-
Chip boom fuels Korea's exceptional growth — but for how long?
South Korea enjoys a boom year in 2026, with Q3 GDP data suggesting annual growth close to 4%. A view of both side of Han RIver from Namsan Tower. AJP Yoo Na-hyun SEOUL, October 27 (AJP) — South Korea’s economy is growing at a pace rarely seen outside a rebound from crisis with third-quarter figures putting annual growth near 4 percent within reach, but experts worry that the window to turn the chip bonanza into broader prosperity is narrowing like sand slipping through an hourglass. The gross domestic product grew 1.0 percent on quarter and 3.7 percent on year in the July–September period, according to preliminary Bank of Korea (BOK) data released Tuesday. Annualized growth over the first nine months would be 3.6 percent. “If fourth-quarter growth reaches 1.4 percent or more on quarter, annual growth of 4 percent would be achievable in theory,” said Lee Dong-won, director general of the BOK’s Economic Statistics Department, at a briefing. Growth near 4 percent would put the economy on a pace last seen in the early 2010s, apart from its 4.1 percent rebound in 2021 from the pandemic recession. South Korea grew 3.7 percent in 2011 and 6.8 percent in 2010 amid stimulus following the global financial crisis. Computer, electronic and optical products accounted for about 60 percent of year-on-year growth, with semiconductors alone contributing roughly half, according to Lee. Edited and compiled by AJP. Graphics AI assisted That concentration was a concern for former BOK governor Rhee Chang-yong and other economists at a conference hosted Tuesday by the Federation of Korean Industries (FKI) and the Peterson Institute for International Economics (PIIE). Rhee, now a nonresident senior fellow at PIIE, said Korea was “focusing too much on the investment in electricity, the hardware,” arguing that education, regulation and labor flexibility could prove more binding constraints on the development of its AI industry. Lee Jae-woo, a visiting research fellow at the Korea Economic Research Institute, described the semiconductor boom as potentially “too much of a good thing.” Korea was becoming a two-track economy in which chips thrive while much of the rest struggles to catch up, he said. The divergence could complicate monetary policy. Chip-driven income and growth strengthen the overall economy, while the burden of tight financial conditions falls more heavily on weaker sectors, Lee said. BOK research suggests that widespread AI adoption has yet to deliver comparable productivity gains. About 51.8 percent of Korean workers used generative AI for work in 2025. Users reduced the time needed to perform tasks by an average of 3.8 percent, equivalent to about 1.5 hours in a 40-hour workweek and implying a potential productivity gain of around 1 percent. Yet the central bank found essentially no relationship between time saved and actual output growth. It attributed the disconnect partly to limited changes in workflows, organizational structures and the division of tasks between workers and AI. Ayhan Kose, deputy chief economist of the World Bank Group, said estimates of AI’s effect on productivity remain highly uncertain, ranging from close to zero to around 1 percentage point in existing research. Sizable gains could help reverse part of a decades-long slowdown in global potential growth, he said. But economies would need investment in infrastructure, human capital, data and connectivity to turn the technology into sustained expansion. AI could also raise worker productivity and create new tasks in health, education and agriculture, Kose said, although evidence on its eventual employment effects remains mixed. Former Bank of Korea Gov. Rhee Chang-yong, center, speaks as Peterson Institute for International Economics President Adam Posen and Korea Economic Research Institute President Chung Chul listen during a panel discussion at the FKI-PIIE Joint International Conference in Seoul on Oct. 27, 2026. AJP Kim Yeon-jae For Korea, the challenge extends beyond building the technology. Martin Chorzempa, Dennis Weatherstone Senior Fellow at PIIE, said the country needs the skills to “map AI capabilities to applications in your economy,” rather than treating an all-out race to develop domestic frontier foundation models as the only route to competitiveness. Developing such large, general-purpose models is extraordinarily costly, he said. Korean companies also need the ability to select suitable models, including open-weight models whose underlying parameters are available for adaptation, and fine-tune them for domestic industries. Continued competition between U.S. and Chinese developers could benefit Korean users by keeping prices down and widening their choices, Chorzempa said. He also saw Korea’s investment in computing infrastructure as a source of strategic flexibility: the same data centers can run models developed in either the United States or China. But that competition carries risks for the industry currently driving Korea’s growth. Intense price competition could make it harder for AI developers to recover the enormous cost of training frontier models, Chorzempa said. If that weakens incentives to keep building data centers, the slowdown could feed through to demand for Korean memory chips. Rhee pointed to a further vulnerability. U.S. restrictions on China’s access to advanced semiconductor technology may have benefited Korean chipmakers in the short term, he said. China’s rapid gains in industries with structures similar to Korea’s could nevertheless pose a more fundamental long-term challenge. Korea’s task is to use the semiconductor windfall to lift productivity across the broader economy before its current advantage narrows. The chip boom is buying time. Turning that time into a more productive AI economy remains unfinished work.