When the Korea Employment Information Service compiled its 2025-2035 outlook for 182 occupations, the 205 occupations minus 23 managerial ones, not a single occupation was classified as 'decline.' 'Status quo' accounted for 114, or 62.6%, followed by 'slight increase' with 47 (25.8%), 'increase' with 9 (4.9%) and 'slight decrease' with 12 (6.6%). It is a picture of a different grain from forecasts that AI will carry off jobs whole. The content is contained in the '2025-2035 Qualitative Job Outlook Project Report.'
The researchers, Kim Dong-kyu, Kim Hyung-rae, Lee Eun-su and Yoon Mi-hee, located the core of the change in job tasks. Their analysis is that AI is likely to work as a complementary technology that takes off repetitive work and raises the weight of problem-solving, creativity and interpersonal service. Job tasks here refer to the bundles of individual duties that make up one occupation. It means the occupation of nurse stays as it is, while the record-keeping work nurses used to do changes.
Responses from companies point in a similar direction. 'The Spread of AI and the Tasks of Labor Market Transition,' which the Korea Labor Institute carried in the April issue of its Monthly Labor Review, analyzed a survey conducted last year by the institute together with a broadcaster and an employers' organization. It covered 110 member companies of the Korea Enterprises Federation and 3,227 workers aged 18 or older nationwide. About half of the companies that had actually used AI answered that it would be difficult to replace more than 70% of workers' duties with AI.
The adoption rate itself is not low. Of the responding companies, 52.7% had introduced AI technology, 15.5% used AI robots as well, and 2.7% used robots only. Businesses using no technology at all were counted at 29.1%. Assessments of productivity came close to unanimous, with 97.4% of user businesses answering positively and 93.8% of non-user businesses expecting the same.
An interesting split appears in the labor cost item. Among workplaces not using AI, 90.6% answered that it would help cut labor costs, while at workplaces that had actually introduced it the figure came down to 82.0%. It means expectations before use run higher than assessments after use. On worker safety and health as well, 68.9% of non-using workplaces expected a positive effect and 21.9% answered that there would be little impact.
Views from outside have also leaned toward wariness of overheating. Bank of America's global research team issued a report on the 28th of last month (local time) and assessed that the 'AI Armageddon narrative' fits neither economic theory nor the evidence. Eighty-five years of labor market data were the grounds. Sixty percent of the jobs now in the United States are in occupations that did not exist in 1940, and agriculture, which employed about 40% of Americans in the early 1900s, is now at the level of 1% of total employment.
Within this report, too, the figures split in two. BofA estimated that some 840 million jobs worldwide, 25% of the total, are within the reach of generative AI, and that 33% are exposed in high-income countries. In the International Labour Organization data cited alongside, by contrast, the 'complement' category where AI raises productivity is 13%, and jobs with a high possibility of actual automation are 2.3%. Jobs affected and jobs replaced are separate categories, and this gap shows the distance between fear and reality as it is.
The ATM case BofA presented came with a rebuttal attached. The report's explanation is that after cash machines spread in the 1970s-80s, bank tellers were redeployed into sales and customer service and teller employment actually rose. David Oks, an economist and columnist, countered that this is a half-story. His point is that teller employment fell sharply as mobile banking spread after the early 2000s.
What working people feel runs at odds with the outlook table. In a survey citing a MetLife report, 80% of U.S. employers had introduced AI into their work and 83% answered that work was being handled faster, but 59% of employee respondents took AI as a job insecurity factor rather than a new opportunity. Experts' 10-year forecasts and workers' sense of today sit on different levels. Nvidia CEO Jensen Huang likewise said that even if AI replaces tasks, roles remain.
In the end, the unit of preparation also comes down to the task list. Which three of the ten things that fill a day are likely to pass over to AI, and what will fill the emptied share, is the practical preparation. The detailed list of the 56 occupations the Korea Employment Information Service report classified as increasing is only now going through disclosure procedures. When that list comes out, it will be worth setting beside one's own task list and reading.
