Stanford economists find AI reshapes work but avoids mass layoffs - DongA Science

Contrary to fears that artificial intelligence will steal jobs, experts say its impact has yet to be clearly observed in the employment market. Provided by Getty Image Bank. Contrary to projections that artificial intelligence (AI) will lead to large-scale job cuts, a new analysis finds there has been no clear evidence of AI-driven unemployment in the labor market so far. Instead, AI proficiency is becoming more important in recruitment, existing roles are being consolidated, and changes in how work is done and what skills are required are emerging ahead of any significant reduction in the number of jobs. According to a report by the UK-based Guardian on the 12th (local time), economics experts say the introduction of AI has not yet led to large-scale job losses. Since last year, concerns over job cuts triggered by AI adoption have spread rapidly. In May last year, Dario Amodei, CEO of Anthropic, predicted that half of all entry-level office jobs could disappear. A month later, Sam Altman, CEO of OpenAI, said some occupations would vanish completely. In reality, some companies have carried out restructuring, citing “AI adoption” as the reason. However, a policy report published last month by the Stanford Institute for Economic Policy Research (SIEPR) found no evidence that AI has yet had a major impact on the job market. The report assigned an “AI exposure level” to each occupation based on how much it is affected by AI. The higher the AI exposure level, the greater the likelihood that an occupation will be affected by AI adoption. According to the report, unemployment among the top 20% of jobs with the highest AI exposure increased by 0.77 percentage points (%p) after the emergence of ChatGPT in 2022. This is actually lower than the 0.85 percentage-point increase seen in the bottom 20% of jobs with the lowest AI exposure. Earlier this year, the unemployment rate for recent U.S. college graduates reached 5.6%, higher than the national average of 4.2%, prompting speculation about AI’s impact. The report suggested that other factors may be at play, such as the expansion of remote work and a correction following over-hiring during the COVID-19 pandemic. Erica McIntyre, a senior research scholar at the Stanford Institute for Economic Policy Research, assessed that “even in occupations where AI’s impact is expected to appear first, employment trends remain broadly stable.” She added, “It took decades for the computer revolution to fully transform the labor market,” and analyzed that “the current situation is similar.” The most visible change at the moment lies not in the number of jobs but in the nature of the work. Instead of hiring new staff for tasks that can be automated, companies are consolidating existing roles and demanding higher productivity and AI skills from job seekers. A survey by U.S. online recruitment platform ZipRecruiter found that about 74% of companies surveyed either considered AI skills a mandatory requirement in hiring or gave them significant extra weight. Half of the companies said they “expect candidates to be able to use AI at or above a practical workplace level from their first day on the job.” ZipRecruiter concluded that, rather than a direct decrease in the number of jobs, there is a clear trend toward higher hiring standards. It also noted that the growing difficulty in finding employment stems less from a simple shortage of jobs and more from a mismatch between the skills companies demand and the capabilities of job seekers. Robert Seamans, a professor at New York University’s Stern School of Business who helped establish a standard framework for measuring AI’s impact on jobs, divides AI’s effects into three categories: jobs that disappear, jobs that are created, and jobs where the way work is done changes. “The third category overwhelmingly dominates,” he said, explaining that “AI, like computers and the internet, will primarily alter how humans work.” In the long term, the very structure of employment may also change. Paul Osterman, emeritus professor at the Massachusetts Institute of Technology (MIT), predicted that as companies figure out what skills they need in the AI era, they are more likely to rely on contract workers or freelancers than to expand full-time staff. “More people will find it difficult to build long-term careers,” he warned. According to Osterman’s research, about 35% of U.S. workers are already in forms of employment that can be relatively easily replaced. “AI will accelerate this trend,” he said. Experts believe it will take considerable time before AI’s full impact on the labor market becomes apparent. Nicholas Bloom, a professor of economics at Stanford University, noted that “hiring new people, changing existing systems, and reorganizing job roles do not happen quickly,” adding that “it will take years for the effects to surface, especially because they are also influenced by political debates over data centers and AI security.” “Not all occupations are in the same situation,” Osterman emphasized, noting that the ability to adapt to change “varies by occupation.” He advised that “high-skilled workers should enhance their capabilities and expand external networks to gain bargaining power in the labor market,” and added that “public policy is needed to protect low-skilled workers.” Copyright ⓒ DongA Science. All rights reserved.