/Tech3h ago

Bank of Korea finds generative AI fails to boost macroeconomic productivity despite 51.8% worker adoption

Story Overview

A Bank of Korea study tracking the first three years of generative AI shows 51.8 percent of Korean workers now use tools like ChatGPT on the job, yet national productivity metrics have barely budged. Workers report only a 3.8 percent average drop in task time—about 1.5 hours weekly—with the freed minutes mostly disappearing into extra feedback rounds and reporting loops rather than higher output. Just 4.4 percent of tasks saw meaningful acceleration.

9377135.1K

#1049

Original post

Rohan Paul@rohanpaul_ai#1257inTech

The Bank of Korea just released a report about AI productivity.

Korean workers using generative AI cut task time by 3.8%, about 1.5hours/week on a 40-hour schedule, yet the link between saved time and more work completed was 0.

Only 4.4% of tasks saw time savings above 20%

This creates an AI productivity disconnect, where faster reports can create more reports, faster review can create more review, and saved time can get absorbed by organizational habits instead of becoming higher output.

12:05 AM · Jun 27, 2026 · 4.1K Views

Open Question

Time cuts stay trapped inside existing workflows

Even when individuals finish pieces faster, the overall volume of work processed does not rise. The study points to rigid approval steps and organizational bottlenecks that swallow any individual gains before they compound.

Productivity Watch

Only high-autonomy roles show any lift

Self-employed professionals and heavy users sometimes convert saved time into extra output, but this pattern stays rare across the broader workforce. Macro data through mid-2026 still shows no visible uptick in GDP per hour worked.

Sentiment

Sentiment building, check back later.

Cluster Engagement

Digg Deeper

No Digg Deeper questions have been answered for this story yet.

Related links

AI 도입은 생산성을 높이는가?

한국은행Via

#1257

Posts from X

Most Activity

VIEWS1.1KBOOKMARKS2LIKES2

Rohan Paul@rohanpaul_ai

https://www.bok.or.kr/portal/bbs/B0000347/view.do?nttId=10098529&searchCnd=1&searchKwd=&depth2=201106&depth=201106&pageUnit=10&pageIndex=1&programType=newsData&menuNo=201106&oldMenuNo=201106

Rohan Paul@rohanpaul_ai

The Bank of Korea just released a report about AI productivity.

Korean workers using generative AI cut task time by 3.8%, about 1.5hours/week on a 40-hour schedule, yet the link between saved time and more work completed was 0.

Only 4.4% of tasks saw time savings above 20%

3h1.1K22

RETWEETS1

William Isaac@wsisaac

I've noticed this trend of AI time savings failing to fully manifest and instead being replaced with repetitive work. Technological affordances also require organizational change in order for the savings to benefit workers.

Rohan Paul@rohanpaul_ai

The Bank of Korea just released a report about AI productivity.

Korean workers using generative AI cut task time by 3.8%, about 1.5hours/week on a 40-hour schedule, yet the link between saved time and more work completed was 0.

Only 4.4% of tasks saw time savings above 20%

1h13111

netrunner@netrunner_btc

@rohanpaul_ai The correlation is basically 0

2h29

Venkat Madala@venky4a

@rohanpaul_ai Maybe metrics maybe re looked into , traditional method don’t work

3h17

Jasper 🌰@building BBX@bbxjasper

@rohanpaul_ai The missing variable is what fills the freed-up hour. If it refills with meetings and Slack, output stays flat on paper. Time saved only converts to throughput when the workflow around it actually changes.

2h4

Moonlit Monkey@MoonlitMonkey69

@rohanpaul_ai Computerization took over a decade to be systemized in business enough for productivity gain to be shown, even though that whole time everyone _assumed_ it was increased already.

Think this will be the same.

3h3

Timur Yessenov@Timur_Yessenov

@rohanpaul_ai That 3.8% time saving with 0.00 correlation to output is the enterprise AI problem in one stat. My read: saved time turns into slack unless the workflow owner changes handoffs, incentives, and sign-offs. I’d ask where the 1.5h/week actually went.