There is a wide gap between the fear that AI is wiping out jobs altogether and the actual statistics. A report released by Boston Consulting Group (BCG) on April 3 projected that 50-55% of U.S. jobs will be transformed or reorganized by AI within two to three years, while the share that disappears entirely will stay at 10-15%. That means the share that disappears is small and the share that changes is large.
Yet hiring indicators at home and abroad over the same period show a sharp contraction, out of keeping with this moderate forecast, concentrated in one place. It is the first gateway into the labor market, entry-level hiring.
BCG did not judge occupations as a whole but broke its analysis into three variables: the possibility of automating tasks, whether they can be replaced, and the scalability of demand. As a result, call center agents were classified on the side of disappearance, and software engineers on the side of changing job content. The calculation is that when development costs fall, the amount of software that can be built increases, so engineers' workloads actually grow. In fact, the number of engineers at AI-focused software companies has grown by an annual average of 6.5% since 2023.
Even though the total is rising, what people feel is the opposite. In a survey of 500 office and managerial workers conducted by the Korea Labor Institute last November, 58% said they could lose their jobs to AI within 10 years. The intensity of the anxiety varied by rank and age.
The share reporting job insecurity above a considerable level was 36.4% among assistant managers and below and 25.5% among general managers and above, and it was 35.2% among those in their 20s and 27.3% among those in their 50s. Those who have just entered an organization are more shaken than those who have held on for a long time.
There is a basis for this feeling. Looking at the distribution of developers by years of experience in Software Policy & Research Institute statistics, only one experience bracket saw its share fall between 2022 and 2024. It was juniors with less than three years of experience, which slid from 26.9% to 20.7%. Other brackets held steady or grew, meaning only the entry segment was hollowed out.
The scale of hiring itself has also shrunk. New hiring across all affiliates of one major domestic IT company fell from 600 people in 2022 to 259 in 2024, less than half. Macro indicators point to the same place.
An examination of microdata from the Korean Statistical Information Service shows that in February this year the number of workers in professional, scientific and technical services and in information and communications was down about 147,000 from a year earlier. It is the first time in five years that both sectors declined at once, and the size of the decline is the largest since the industrial classification was revised in 2013.
Who declined is the key. About 89% of the decrease, or 131,000 people, came from those in their 20s (97,000) and 30s (34,000). The distribution looks closer to the result of halting new hiring than to traces of pushing out experienced workers. There is as yet no basis for sorting out how much of this decline is due to AI adoption and how much to a weak business climate.
The shift in perception at overseas companies is also clear. In a survey Accenture conducted in February and March in the U.K. and Northern Ireland covering 2,085 workers and 510 executives, about 50% of executives expected net employment across the economy to fall. In the same survey two years ago, the figure was around one third.
Responses saying AI would increase demand for entry-level workers dropped from 40% to 15%, and responses saying it would decrease jumped from the 20% range to close to 40%. Over the same period about 20% of U.K. workers came to use generative AI daily, three times the level of two years ago.
That does not mean a collapse in the total is in store. The International Labour Organization (ILO) projected that the global unemployment rate will hold at the 4.9% mark this year. Instead the burden falls on young people.
Last year the global youth unemployment rate was 12.4%, and the share belonging to neither education, employment nor training was around 20%. The ILO judged that companies, unable to calculate the effects of AI adoption, are holding back on new hiring, and that the burden is concentrated on young people at the entry stage.
From a company's standpoint, entry-level hiring is an investment that pays off late. With AI handling junior-level tasks instead, the choice to defer that investment looks rational in the short term. Accenture analyzed 17 industries and said AI can generate revenue more than twice the labor costs it can save, naming energy, life sciences and retail as beneficiary fields. BCG likewise cautioned that using AI only as a cost-cutting tool lowers productivity and drives talent away, recommending that retraining and the redesign of work processes go together.
Some three years after generative AI went mainstream at the end of 2022, the mark left on the labor market is a narrowed entry point more than mass layoffs. Neither companies nor the government has produced an answer to where the next generation will build skills once machines handle junior-level work, or where mid-career staff will come from five years on. How the scale of entry-level hiring is set in the open recruitment season in the second half of this year becomes the first indicator for gauging whether this trend continues. For someone looking for a first job, AI is felt less as a tool than as the height of a threshold to be crossed.
