Seniority-Biased Technological Change

Definition and Origin

The term comes from the observation that generative AI adoption acts differently by career stage even within the same occupation. Seyed M. Hosseini and Guy Lichtinger of Harvard posed the research question in their paper "Generative AI as Seniority-Biased Technological Change: Evidence from U.S. Résumé and Job Posting Data," first released on 31 August 2025, of whether generative AI constitutes a form of seniority-biased technological change that disproportionately affects junior relative to senior workers. The paper uses U.S. résumé and job posting data covering nearly 62 million workers in 285,000 firms (2015–2025), identifies generative AI adoption via job postings flagging dedicated "AI integrator" roles, and uses difference-in-differences and triple-difference estimates. Junior employment in adopting firms declined sharply from 2023Q1 relative to non-adopters while senior employment continued to rise. The paper was presented at the Stanford Digital Economy Lab seminar on 22 September 2025.

Adoption into Korean Policy Discourse

The Bank of Korea introduced the concept into Korean policy discourse in BOK Issue Note No. 2025-30 (30 October 2025), "AI 확산과 청년고용 위축: 연공편향(seniority-biased) 기술변화를 중심으로" (AI diffusion and the contraction of youth employment: focusing on seniority-biased technological change), using National Pension Service subscriber data of roughly 16 million subscribers as large-scale administrative statistics. It was popularized by the Bank's blog post of 6 November 2025, "AI 확산 초기, 청년고용은 왜 감소하는가?" (Why does youth employment fall in the early phase of AI diffusion?). The Bank explicitly compares its finding with the U.S. result ("similar to the United States"). The Bank writes the term as "연공편향(seniority-biased) 기술변화" and defines the phenomenon of falling junior employment alongside rising senior employment in a single sentence.

Mechanism

According to the Bank of Korea, AI readily substitutes for the codified knowledge tasks and book-learning knowledge that juniors predominantly handle, while acting as a complement to seniors' tacit knowledge, organizational management, contextual understanding and social skills. Among junior workers with up to five years of tenure, four-year-college bachelor's and master's holders showed the largest AI-related reduction in worktime, with a U-shaped pattern by education.

Korean Evidence and Figures

From July 2022 to July 2025, youth (15–29) jobs fell by 211,000, of which 208,000 (98.6%) were lost in AI high-exposure industries. Over the same period, jobs for workers in their 50s grew by 209,000, of which 146,000 (69.9%) were in high-exposure industries. Industry-level youth employment declines were −11.2% in computer programming and systems integration/management, −8.8% in professional services, −20.4% in publishing, and −23.8% in information services. However, in industries with high AI exposure but high complementarity (health, education services, air transport), youth employment decline was small or absent, and wage effects were not yet clear.

U.S. Evidence

In the United States, the junior employment decline was concentrated in the most generative-AI-exposed occupations and was driven primarily by slower hiring rather than increased separations. Education heterogeneity showed a U-shaped pattern, with mid-tier graduates most affected and elite and low-tier graduates less affected.

Distinctions

Classic skill-biased technological change (SBTC), a concept associated with Katz and Murphy (1992), holds that technological change raises the productivity of skilled workers relative to unskilled labor, and it has been invoked to explain the skill premium and wage inequality. Seniority-biased technological change shifts this axis from skill and education to career stage (tenure) within the same occupation. The Bank of Korea's finding of a U-shaped education pattern among juniors, in which mid-tier graduates are the most substitutable, cuts against a pure skill-level story. The attribution of the SBTC lineage, however, rests on search-snippet-level leads (Katz & Murphy 1992; Violante 2008; Acemoglu; Card) rather than a fetched primary source, and needs author verification.

A further disambiguation point is that in English both seniority-biased technological change and skill-biased technological change circulate under the acronym SBTC.

Relations and Measurement

Seniority-biased technological change is a species of technological change. In economics, technological change is defined as a change in the set of feasible production possibilities, classified into Hicks-neutral, Harrod-neutral and Solow-neutral forms, and classic skill-biased technological change is one such biased form. The Bank of Korea measures AI exposure and complementarity using Felten et al. (2021) and Pizzinelli et al. (2023), relating the concept to the AI-exposure and AI-complementarity literature. It also cross-checks the effect against household-survey AI usage rates from BOK Issue Note 2025-22 and regression controls.

Contested Points and Uncertainty

The Bank of Korea dates the onset to the release of ChatGPT in November 2022 and describes its finding as the "early phase" of AI diffusion. The U.S. estimates begin in 2023Q1. The Bank states explicitly that it is uncertain whether the youth employment contraction will persist, noting that firms may shift to sustainable talent-development strategies and that long-run effects on career paths and income inequality warrant continued attention.

A Korean press account reports the gain for workers in their 50s as 209,000, with about 70% (146,000) in high-exposure industries, and lists −11.2% (programming), −20.4% (publishing), −8.8% (professional services) and −23.8% (information services). Because the ordering of 20.4% and 8.8% differs from the Bank's blog, the Bank's blog is treated as primary for source-ordered figures. The same account attributes substitution pressure to finance and insurance and some office support services, and complementarity to R&D, consulting and legal services, and reports the Bank's policy recommendation to shift firms from seniority-based toward performance- and job-based hiring and compensation.

Naming

The term circulates in English as "seniority-biased technological change" and in Korean as "연공편향 기술변화" and "연공편향(seniority-biased) 기술변화." The Bank of Korea and the press also use "연공편향적 기술변화," and the paper title uses "generative AI as seniority-biased technological change."

Sources

• Bank of Korea BOK Issue Note No. 2025-30 (30 October 2025), "AI 확산과 청년고용 위축: 연공편향(seniority-biased) 기술변화를 중심으로"
• Bank of Korea blog post (6 November 2025), "AI 확산 초기, 청년고용은 왜 감소하는가?"
• Hosseini, Seyed M. & Guy Lichtinger, "Generative AI as Seniority-Biased Technological Change: Evidence from U.S. Résumé and Job Posting Data" (2025)

Notes

• The English acronym SBTC collides with skill-biased technological change.
• The Bank of Korea compares its result with the U.S. result as similar.
• The Bank itself flags uncertainty about causality and persistence.
• The SBTC lineage attribution (Katz & Murphy) needs primary-source verification.