SEOUL – On a Tuesday night in April, seven office workers gathered around a table in a co-working space in southern Seoul, each having paid 70,000 won ($50) for a 90-minute, hands-on lesson on Claude Code, an agentic coding tool developed by U.S. AI company Anthropic.
Ok Ye Jin, 33, was among them. She works for a direct-to-consumer online store of a multinational brand’s Korean unit. Ok said she signed up voluntarily to learn how to use the smart assistant, which designs computer programs based on users’ descriptions of the features they want in natural language.
“My company’s security policy does not allow us to use most outside AI tools in the workplace,” she said, explaining that she had come on her own time and at her own expense to learn one anyway.
The instructor, Kim Suha, a 33-year-old corporate strategy and investment professional, teaches her students how to interact comfortably with the AI agent to make their jobs — and ultimately their lives — easier as the technology advances.
Her five sessions, each with only seven seats available, sold out almost immediately after she announced them. This is part of a broader trend of South Korean office workers teaching themselves AI outside of work, often without encouragement from their employers.
A recent study by Bank of Korea researchers revealed a striking gap in the adoption of generative AI between South Korean workers and their employers, with employees using tools that many companies have not yet formally approved.
According to a separate government survey cited in the study, only 6.2 percent of mid- to large-sized firms had formally adopted generative AI as of 2023. However, at the time of the BOK survey in mid-2025, 51.8 percent of workers reported using such tools on the job, and 78.6 percent of those users spent more than an hour a day on them.
Aware of the increase in interest, the South Korean education ministry announced in February that it would expand the number of university-run AI training programs for working adults to 38 institutions this year, up from 30 in the program’s first year, which drew nearly 12,000 participants.
Huh Da Eun, a 34-year-old investment analyst at a South Korean venture capital firm, agreed that office workers need formal training. Her workplace does not offer this, even though it encourages staff to use generative AI for routine tasks, which is why she too paid for Kim’s course.
Like most people, Kim had assumed that any task involving code was something she had to ask a developer to handle.
However, she realized this was not the case when, late last year, she watched her husband — who is also not a programmer — use ChatGPT to analyze TikTok hashtag data in just minutes.
She then started using Claude Code to access different websites to collect reviews for her job, a task which was completed in a fraction of the time she had spent gathering them herself. “That small victory was the starting point,” she said.
The biggest change since then, she said, has been in the way she approaches work itself. “I feel like I’ve moved from the perspective of a doer to that of a planner,” Kim said.
Her transformation from student to teacher came in March after spending a week using Claude Code, which she described as her most significant turning point in recent years.
Friends and acquaintances began asking her to show them how it works, and those requests eventually led to the five sold-out sessions.
Also at the table that evening was Park Jeong Ah, a 33-year-old content marketer whose company regularly holds in-house AI training sessions for its staff. She said that even those programs had left her wanting to learn more, which is why she enrolled in an external session.
Not everyone, however, is eager to learn the technology.
In May 2025, Now & Survey conducted a survey of 1,000 South Korean office workers and found that 27.3 percent said they felt stressed about keeping up with generative AI at work.
However, the same survey showed that, on a 10-point scale, respondents’ overall sense of optimism about the technology stood at 7.1, which is still higher than their anxiety score of 5.9.
Ok, the e-commerce manager, said that anxiety was what had pushed her to sign up.
“I kept feeling like I was going to fall behind if I didn’t learn to use AI well,” she said.
Kim said that, within days, most of her students end up with the same set of questions: Is the technology dangerous? Are their jobs about to disappear? And how can they turn it into a career asset before the answers to these questions become obvious?
Kim said that she never intended to run a business and does not plan to hold more sessions. She wants to return to her own AI projects and push her company toward what she terms “AX,” or AI transformation.
She said that what drives her is an image she obtained from an AI camp she once attended: a highway full of self-driving cars capable of traveling 500 kilometers per hour, except for one human driver who insists on driving himself.
“Because of that one car, everyone else has to slow down to 100,” Kim said.
“I want to be in a place where everyone is running at the full speed the machine allows,” she added.
