AI in Everyday Life: How Workers Use AI

Research Department, Labor Market Research Team

저자: Economist Donghyun Suh, Head Samil Oh, Junior Economist Minjeong Kim

2026.01.06

키워드: AI Labor Market Productivity Household Survey

첨부파일

(2026.1.6.) AI in Everyday Life_How Workers Use AI.pdf
Download Viewer

Generative AI is no longer just something we hear about in the news. It has already become a part of everyday life—used for brainstorming new ideas, language translation, content creation, and technical analysis in a wide range of areas. Only a few years ago, many people saw it as a technology reserved for experts. Today, it has become a tool that almost anyone can use, at work and at home. So how often—and in what ways—are people actually using AI? And has it actually improved productivity? To find out the answers to these questions, we conducted Korea’s first representative survey^[1] on the use of generative AI^[2].

63.5% of workers have used generative AI
Adoption is twice the U.S. level and eight times faster than the Internet

Our survey shows that 63.5% of Korean workers have used generative AI at least once. Even when we focus only on work-related use, the share still exceeds half of all workers (51.8%)[Figure 1]. What really stands out is the speed of adoption: the rate is about twice the U.S. figure (26.5%) and roughly eight times the Internet adoption rate three years after commercialization (7.8%)[Figure 2]. Why has generative AI spread so quickly? Korea already has a strong digital foundation—built around the Internet and smartphones. On top of that, generative AI is a truly general-purpose technology, which makes it useful across a wide range of tasks and occupations.

Figure 1. Generative AI Adoption in Korea and the U.S.^[3]

A chart comparing generative AI adoption rates in Korea and the U.S., separately for overall use, use for work, and use outside of work. 1. Overall: Korea 63.5%, United States 39.6% 2. For work: Korea 51.8%, United States 26.5% 3. Outside of work: Korea 60.1%, United States 33.7%

Figure 2. The Trajectory of Generative AI¹ and the Internet² Adoption

A chart showing adoption rates over the years since commercialization for generative AI and the Internet. 1. Generative AI reached an adoption rate of 63.5% in under three years (since 2022). 2. The Internet reached an adoption rate of 7.8% three years after commercialization.

Higher adoption among men, younger workers,
and those with higher income and education

Generative AI is now used across the labor market, but adoption varies by personal characteristics and occupation. Men (55.1%) report higher usage than women (47.7%)[Figure 3]. Among young adults (ages 18~29), adoption reaches 67.5%—about twice the rate among older workers (ages 50~64)[Figure 4]. Adoption also tends to rise with income and education[Figure 5,6]. For example, the adoption rate is 72.9% among those with a graduate degree, compared with 38.4% among those with less than a college degree.

Figure 3. AI Adoption by Gender

A chart showing generative AI adoption rates for men and women, broken down into “used every day in the past week,” “used at least once in the past week,” and “did not use in the past week.” 1. Overall adoption: Men 55.1%, Women 47.7% 2. Men: 7.6% used every day in the past week; 33.6% used at least once in the past week; 13.9% did not use in the past week. 3. Women: 5.5% used every day in the past week; 27.2% used at least once in the past week; 15.0% did not use in the past week.

Figure 4. AI Adoption by Age Group

A chart showing generative AI adoption rates by age group, broken down into “used every day in the past week,” “used at least once in the past week,” and “did not use in the past week.” Overall adoption: 18–29: 67.5%; 30–39: 65.3%; 40–49: 55.1%; 50–64: 35.6% Ages 18–29: 9.5% used every day; 41.3% used at least once; 16.7% did not use in the past week. Ages 30–39: 10.1% used every day; 39.6% used at least once; 15.6% did not use in the past week. Ages 40–49: 6.9% used every day; 32.7% used at least once; 15.5% did not use in the past week. Ages 50–64: 3.3% used every day; 20.2% used at least once; 12.1% did not use in the past week.

Figure 5. AI Adoption by Income Level

A chart showing generative AI adoption rates across income percentiles. Adoption generally increases as income rises.

Figure 6. AI Adoption by Education Level

A chart showing generative AI adoption rates by education level, broken down into “used every day in the past week,” “used at least once in the past week,” and “did not use in the past week.” 1. Overall adoption: Graduate 72.9%; Bachelor’s 59.8%; Less than college 38.4% 2. Graduate: 11.7% used every day; 43.9% used at least once; 17.3% did not use in the past week. 3. Bachelor’s: 7.9% used every day; 36.7% used at least once; 15.2% did not use in the past week. 4. Less than college: 4.1% used every day; 21.5% used at least once; 12.8% did not use in the past week.

5~7 hours of AI use per week,
higher intensity than in the U.S.

Among workers who use generative AI, the reported time spent using AI is 5~7 hours per week—equivalent to 12.1%~16.6% of a 40-hour workweek. That is a much higher level of AI use intensity than in the United States (0.5~2.2 hours per week). The share of heavy users—those who use AI for at least one hour per day—is also much higher in Korea (78.6%)[Figure 7] than in the United States (31.8%)[Figure 8].

Figure 7. AI Use Intensity in Korea

A chart showing the intensity of AI use in Korea as shares by average daily time spent using AI. 1. All workers: 78.6% use AI for 60 minutes or more per day; 17.3% for 15–59 minutes; 4.1% for 0–14 minutes. 2. Workers who used AI at least once in the past week: 76.1% for 60 minutes or more; 19.4% for 15–59 minutes; 4.5% for 0–14 minutes. 3. Workers who used AI every day in the past week: 90.2% for 60 minutes or more; 7.5% for 15–59 minutes; 2.3% for 0–14 minutes.

Figure 8. AI Use Intensity in the U.S.

A chart showing the intensity of AI use in the U.S. as shares by average daily time spent using AI. 1. All workers: 31.8% use AI for 60 minutes or more per day; 47.1% for 15–59 minutes; 21.1% for 0–14 minutes. 2. Workers who used AI at least once in the past week: 22.6% for 60 minutes or more; 54.0% for 15–59 minutes; 23.4% for 0–14 minutes. 3. Workers who used AI every day in the past week: 51.9% for 60 minutes or more; 37.1% for 15–59 minutes; 11.1% for 0–14 minutes.

AI use reduces work hours by 3.8% and raises potential productivity by 1.0%

After adopting generative AI, workers’ average weekly work hours fell by 3.8% (about 1.5 hours per week under a 40-hour schedule). This translates into an estimated 1.0% gain in potential productivity^[4]. The estimate is similar to the figure reported by Bick et al. (2025) for the United States (1.1%) and meaningfully higher than Acemoglu’s (2024) projection (0.7% over the next ten years)[Figure 9].

Figure 9. Potential Productivity Gains

A chart comparing estimated productivity gains across Korea, the United States, and Acemoglu (2024): Korea 1.0%, United States 1.1%, Acemoglu (2024) 0.7%.

The effects also differ across occupations. The largest reductions in work hours (1.5%~2.8%) are observed among managers and professionals, while the effects are relatively smaller in occupations centered on physical labor[Figure 10]. Reductions are also larger for less-experienced workers, suggesting an equalizing effect—generative AI may help narrow gaps in task proficiency[Figure 11]. Still, 54.1% of AI users report no reduction in working hours. One reason may be that some workers are still getting used to these tools. Another possibility is that reviewing and verifying AI-generated outputs takes additional time. As more workers learn to use generative AI effectively—and as the technology becomes more reliable—productivity gains could become larger over time.

Figure 10. Reduction in Work Hours by Occupation¹

A chart showing the percentage reduction in work hours by nine major occupational groups. Professionals and related workers: 2.8%, Clerical support workers: 1.9%, Managers: 1.5%, Skilled agricultural, forestry and fishery workers: 0.9%, Plant and machine operators and assemblers: 0.9%, Sales workers: 0.7%, Service workers: 0.6%, Craft and related trades workers: 0.5%, Elementary occupations: 0.2%

Figure 11. Regression Analysis of AI-Induced Working Time Reduction¹

A chart showing regression results for work-hour reductions, using the following variables: male; age 20 or younger; 30s; 40s; college graduate; master’s; PhD; income; assets; work motivation; tenure (10 years); work hours (10 hours); physical labor; core tasks; sense of accomplishment at work. Red lines indicate 95% confidence intervals. The reduction in work hours is larger for less-experienced workers.

Nearly half view AI’s impact positively;
many are preparing through training and job transitions

Nearly half of workers (48.6%) believe AI will bring positive changes to society, far exceeding the share expressing negative views (17.5%)[Figure 12]. Meanwhile, 33.4% of workers plan to pursue education or training in response to AI advances, and 31.1% report preparing for a job change[Figure 13]. These preparations are especially common among those who have used generative AI and among workers who expect a high likelihood of automation in their jobs.

Figure 12. Social Impacts of AI Technology

A chart showing how all employed persons and those who use AI for work assess AI’s impact on society. 1. Positive: All employed persons 48.6%; AI-for-work users 53.6% 2. Negative: All employed persons 17.5%; AI-for-work users 17.9% 3. No impact: All employed persons 4.3%; AI-for-work users 4.8% 4. Not sure: All employed persons 29.6%; AI-for-work users 23.7%

Figure 13. Reskilling and Career Change Plans to Adapt to AI-Induced Job Market Changes

A chart showing shares of workers who plan to pursue training or education and who plan to change jobs in preparation for advances in AI. 1. Plan to pursue education/training: 33.4% 2. Plan to change jobs: 31.1%

The need to continuously expand AI-related data

This survey goes beyond simply asking, “Who uses AI, and how much?” It offers one of the first empirical assessments in Korea of how AI is changing work and influencing productivity. Going forward, systematic data collection and analysis on AI use should continue—so that policy can be grounded in stronger evidence and designed with greater precision.

[1] The survey was conducted from May 19 to June 17, 2025 (about one month) and covered 5,512 employed persons aged 15~64 nationwide. The sample was designed based on the Ministry of Data and Statistics’ Local Area Labour Force Survey (the first half of 2024), taking into account occupation, age, and gender.
[2] For further details, BOK Issue Note No. 2025-22, “Rapid Adoption of Artificial Intelligence and Its Productivity Effects: The Case of Korea.”
[3] At present, the United States is one of the very few countries with systematic, representative-sample evidence on generative AI adoption. To facilitate comparison, the survey questionnaire design drew heavily on Bick et al. (2025).
[4] For example, Korea’s GDP grew by 3.9% from Q4 2022 (release of ChatGPT) to Q2 2025. Theoretically, 1.0 percentage point of this growth can be interpreted as the potential contribution of generative AI adoption. However, this figure assumes that workers did not use the saved time for leisure but instead for additional productive activity; if some of the saved time was allocated to leisure, the realized productivity gain would be smaller.

List

Blog