In May, regular employees among South Korea's wage workers numbered 16.74 million, down 7,000 from a year earlier. It is the first decline in 26 years and five months, since December 1999. The trend, which had not been broken once in the 316 months since it turned to growth in January 2000, stopped here. In April, the month just before, the figure was still up 62,000.
Regular employees are workers expected to stay on the job for a year or more. Permanent staff and open-ended contract workers make up the core, and fixed-term contract workers are also included depending on the conditions. Unlike temporary and daily workers, these are jobs seen as continuing, so a drop in this figure means the sturdiest layer of the labor market has been shaken.
The decline is heavily concentrated by age group. Regular jobs fell by 164,000 among people in their 20s and by 34,000 among those in their 30s, a combined loss of 197,000. It is the largest decline since December 2020. Among people in their 20s, temporary jobs fell by 67,000 and daily jobs by 12,000 along with regular jobs.
Looking at industries makes the picture clearer. Regular jobs for people in their 20s fell by 57,000 in information and communications, and for those in their 30s the largest drop was 76,000 in professional, scientific and technical services. Regular manufacturing jobs among people in their 20s and 30s fell by 92,000 as well, and total manufacturing employment fell by 140,000, a 23rd consecutive month of decline. By region, Seoul fell by 39,000 and Gyeonggi by 45,000, while South Gyeongsang rose by 37,000 and Busan by 15,000.
One interesting point is that in the same information and communications industry, regular jobs for people in their 30s rose by 26,000. A pattern of people in their 20s leaving and those in their 30s coming in appeared at once within a single industry. Pinning down the net change for the industry as a whole would require a finer count, but at least the door to this industry has not closed equally for every age.
Overseas data points the other way. According to a PwC report analyzing more than 1 billion job postings worldwide, productivity at companies in industries with high AI exposure rose 34% in 2025 compared with 2018, ahead of companies with low capability to use it (24%), and the average labor productivity growth rate at the top 20% of companies reached 163%. Headcount growth was also higher at companies with high AI exposure, at 52%, above the 36% at companies with low exposure.
Postings requiring automation engineering, machine learning and the like have risen 69% since 2019, about eight times the 8.6% growth rate of the hiring market as a whole. These figures cannot separate out whether adopting AI increased hiring, or whether companies that were doing well simply brought in AI ahead of others.
The grounds for pessimism are considerable too. Goldman Sachs has analyzed that AI is eliminating about 16,000 jobs a month in the United States, with junior-level workers and Generation Z hit hardest. Global tech-sector layoffs counted through May of this year topped 115,000, approaching last year's annual record, and Meta, Amazon and Snap cited AI as one of the reasons for the cuts. In a US opinion poll released in June, half of respondents said AI could lead to job loss for themselves or a family member.
Jeff Bezos, in conversation with Blue Origin's chief executive at a technology conference held in Paris, said he does not agree with the view that AI makes humans redundant. His argument is that labor will instead be in short supply. In an interview last month he also compared a bulldozer with a shovel and said AI raises workers' capabilities rather than replacing them. Anthropic's Dario Amodei, meanwhile, predicted that a painful disruption could come across white-collar work and later took back part of it.
The data from the two camps are in fact not measuring the same thing. PwC looks at the company level, while Goldman Sachs and the layoff tallies look at the United States as a whole and at the tech sector. It is entirely possible for a company to hire more people while cutting entry-level hiring. The National Data Office also takes the position that it is hard to state definitively how much AI has affected this hiring slowdown.
While total employment fell by 40,000, the share of regular jobs hit a record high of 57.5%. It means employment has become more stable for those who remain, and it also means the passage in for newcomers has narrowed by that much. Since the population in its 20s is itself shrinking, the entire decline cannot be blamed on technology.
What changes right away is the route to a first job. The number of jobs itself may stay the same. As the repetitive work that a new hire would once have spent 1-2 years learning moves over to tools, companies have begun looking for people who have skipped that stage.
If hiring that asks what someone has done after four years of university increases, where that experience is to be built is a question companies, schools and the government have to answer together. Whether the first decline in 316 months is a one-month wobble or a trend will be confirmed in the next employment trends report due in July.
