Artificial Intelligence and the Labor Market: Examining the Correlation withYouth Unemployment in Japan

I. Introduction
When OpenAI introduced ChatGPT in November 2022, artificial intelligence (AI) began to drastically reshape society — particularly the labor market. Menial jobs are susceptible to substitution, while jobs requiring empathy or interpersonal skills may be harder to replace.
The fallout is already measurable. Research from Stanford University found that workers aged 22–25 in highly AI-exposed U.S. occupations experienced a 13% decline in employment since 2022 (Atkinson and Yamco, 2026).
Japan offers a useful test case, where an aging population and labor shortages mean AI could potentially fill gaps instead of displacing workers. It examines the correlations for early-stage automation and generative AI, compares Japan with the United States, and considers policies to help young workers adapt.
II. The Japanese Labor Market
Japan’s labor market has traditionally been shaped by lifetime employment and seniority-based promotion (Okina, 2024), yet it faces labor shortages. In 2024, approximately half of surveyed firms reported insufficient qualified full-time employees.

Adoption of the newest technology remains uneven. Only 23.5% of Japanese SMEs used generative AI in 2025 – the lowest among surveyed OECD countries. SMEs account for 99.7% of Japanese businesses and 70% of the workforce (Dyvik, 2025); low adoption is partly attributed to a shortage of AI talent and cost concerns.
Japan’s culture of “simultaneous mass recruiting of new graduates” once supported smoother transitions into work. After the asset bubble burst in the 1990s, many young people struggled to find stable jobs and became part-time workers known as freeters. Japan’s youth unemployment rate reached 3.6% in April 2026.
III. The Adoption of Automation and AI on Japan’s Youth Unemployment
A. Methodology
Because generative AI emerged at the end of 2022, this study also examines early-stage automation. It uses robot density per 10,000 Japanese manufacturing workers, generative AI adoption data, and youth unemployment rates from World Bank Open Data. The study measures the relationship with Pearson’s correlation coefficient; a p-value below 0.05 indicates statistical significance.
B. Correlation of automation and generative AI and Japanese youth unemployment rates

Japanese youth unemployment generally declined, despite rising between 2018 and 2020 and plateauing at 4.7% in 2021, likely due to the COVID pandemic. It reached 3.9% in 2025. Meanwhile, robot density increased steadily, nearly doubling by 2024 compared with 2015, while generative AI adoption surged between 2024 and 2025. The sharp rise in reported adoption may also reflect differences between the sources used to measure AI adoption.


Robot density shows a strong negative relationship with Japanese youth unemployment (r = −0.8901; R² = 0.7923). Its p-value of 0.0072 is below 0.05, making the relationship statistically significant. Generative AI shows a moderate negative relationship (r = −0.5519), but its p-value of 0.6278 is not statistically significant, likely due to the small number of data points.
Overall, both early-stage automation and generative AI may have a negative relationship with youth unemployment in Japan, but only the automation result is statistically significant. Falling unemployment may also reflect demographics, such as an aging population or labor shortage, rather than AI alone.
There are only three data points for generative AI, since it only started in the fall of 2022, when ChatGPT was first established. Hence, the results for the Pearson correlation coefficient are indicative rather than conclusive, as there is not sufficient data. Additionally, RIETI measures worker-level AI usage, while GMO Research measures generative AI usage among working-age individuals, which may result in inconsistencies.
IV. Comparative Analysis with the United States

In the U.S., the 2020 youth unemployment figure is an outlier and is removed from the correlation test. Unlike Japan, U.S. youth unemployment began to rise after 2022. Robot density and generative AI adoption rose in both countries, but Japan had higher robot density while the U.S. had higher generative AI adoption.


Based on the results above, early-stage automation is negatively associated with youth unemployment in both countries, while generative AI shows a negative association in Japan and a positive one in the U.S.
The differences in how generative AI correlates to youth unemployment in Japan and the U.S may not be attributed to the sources of underlying generative AI technology they are using, as they all have access to predominant models like ChatGPT, Claude, or Gemini, but may be highly likely due to the disparities between the dynamics of their labor markets.
The aforementioned fact that Japan has a very distinct corporate culture of long-term employment practices and the severe labor shortage due to the declining birthrate and rise in the aging population acts as a strong buffer, which makes it unlikely for the rate of youth unemployment to rise, unlike the situation in the U.S.(Fukuda, 2026). The U.S, on the other hand, faces a continuous decline in employment in highly exposed AI jobs, as mentioned before, and has a more flexible and less protected labor market structure, in which firms could displace their workers much more easily, causing young workers to be much more susceptible to AI-driven unemployment (Wang, 2026).
V. Conclusion and Policy Recommendations
The results suggest a negative correlation between early-stage automation and youth unemployment in both countries. Generative AI shows a negative correlation in Japan and a positive one in the U.S., but these results are inconclusive due to limited data. The comparison suggests that the relationship between AI and youth unemployment may depend on labor-market structure.
The policy implications follow from that. Governments should consider establishing an online system for AI-related training for young workers to ensure convenience and job security, which involves gathering AI specialists who teach them how to incorporate AI into their daily work to increase their efficiency. Additionally, governments should encourage higher education institutions to cultivate positive personality characteristics or soft skills other than technical knowledge, as these skills are less likely to be acquired by AI. Lastly, as AI would replace certain tasks, policies that target improving labor mobility become crucial. As an example, Japan’s 2023 “New Trinity Reform,” which includes a mixture of investments in retraining programs for reskilling, a shift towards job-based pay from seniority-based payment, and targeting inter-company mobility, effectively addresses this issue (Zou, 2024).
References
Asao, Kohei, Haruki Seitani, Ara Stepanyan, and TengTeng Xu. 2025. “The Impact of Aging and AI on Japan’s Labor Market: Challenges and Opportunities.” IMF Working Paper 2025, no. 184 (September). https://doi.org/10.5089/9798229024747.001.a001.
Atkinson, Tyler, and Shane Yamco. 2026. “Young Workers’ Employment Drops in Occupations with High AI Exposure.” Federal Reserve Bank of Dallas, January 6, 2026. https://www.dallasfed.org/research/economics/2026/0106.
Bick, Alexander, Adam Blandin, and David J. Deming. 2025. “The State of Generative AI Adoption in 2025.” On the Economy Blog, Federal Reserve Bank of St. Louis, November 19, 2025. https://www.stlouisfed.org/on-the-economy/2025/nov/state-generative-ai-adoption-2025.
Canada, Innovation, Science and Economic Development. 2025. SME AI Adoption Blueprint. December 10, 2025. https://ised-isde.canada.ca/site/ised/en/sme-ai-adoption-blueprint.
Dyvik, Einar H. 2025. “SMEs in Japan – Statistics & Facts.” Statista, December 17, 2025. https://www.statista.com/topics/13372/smes-in-japan/.
Federal Reserve Bank of St. Louis. 2026. “Youth Unemployment Rate for the United States [SLUEM1524ZSUSA].” FRED. Accessed July 9, 2026. https://fred.stlouisfed.org/series/SLUEM1524ZSUSA.
Frey, Carl B., and Michael A. Osborne. 2017. “The Future of Employment: How Susceptible Are Jobs to Computerisation?” Technological Forecasting and Social Change 114: 254–280. https://doi.org/10.1016/j.techfore.2016.08.019.
Fukuda, Naoyuki. 2026. “Generative AI and Employment in Japan ― How to Ensure Young People Develop Skills ―.” Japan Research Institute, June 15, 2026. https://www.jri.co.jp/en/reports/jrirj/2026/fukuda/11/.
GMO Research & AI. 2025. “Japan Generative AI Tracking Survey (Feb 2024–Feb 2025).” GMO Research, 2025. https://www.gmo-research.ai/en/resources/studies/2025-study-gen-AI-jp.
Hori, Yukie. 2020. “Japan’s ‘Employment Ice-Age Generation’ Today: Investigating the Impact of Instability in the School-to-Work Transition.” Japan Labour Issues 4, no. 20 (January).
International Federation of Robotics. n.d. “IFR Press Releases.” Accessed [July 1, 2026]. https://ifr.org/ifr-press-releases.
Kim, Hyun-Cheol. 2026. “AI’s Experience Bias Shocks Labor Market, Hitting Unemployed Youth First.” Korea JoongAng Daily, March 10, 2026. https://www.koreajoongangdaily.com/opinion/ais-experience-bias-shocks-labor-market-hitting-unemployed-youth-first/12554004.
Morikawa, Masayuki. 2025. “The Impact of Artificial Intelligence on Macroeconomic Productivity.” RIETI, January 14, 2025. https://www.rieti.go.jp/en/columns/v01_0218.html.
OECD. 2025. Artificial Intelligence and the Labour Market in Japan. Paris: OECD Publishing. https://doi.org/10.1787/b825563e-en.
Okina, Yuri. 2024. “Japan’s Jobs Market on the Way to Modernisation.” East Asia Forum, July 7, 2024. https://eastasiaforum.org/2024/07/07/japans-jobs-market-on-the-way-to-modernisation/.
Schneider, Todd, Eric M. Pondi Endengle, and Anh Van Le. 2018. “Land of the Rising ROBOTS.” Finance & Development 55, no. 2 (June). https://doi.org/10.5089/9781484357415.022.a010.
Trading Economics. n.d. “Japan Youth Unemployment Rate.” Ministry of Internal Affairs & Communications. Accessed [July 1, 2026]. https://tradingeconomics.com/japan/youth-unemployment-rate.
U.S. Bureau of Labor Statistics. 2024. “The Employment Situation: November 2024.” News release, December 6, 2024. https://www.bls.gov/news.release/archives/empsit_12062024.pdf.
U.S. Bureau of Labor Statistics. 2025. “Job Openings and Labor Turnover Survey: March 2025.” News release, April 29, 2025. https://www.bls.gov/news.release/archives/jolts_04292025.pdf.
Wang, Fangyan, et al. 2026. “Generative AI and the Reorganization of Labor Demand.” arXiv, May 2026. https://arxiv.org/abs/2605.23159.
World Bank. n.d. “Unemployment, Youth Total (% of Total Labor Force Ages 15–24) (National Estimate) – Japan.” World Bank Open Data. Accessed [July 1, 2026]. https://data.worldbank.org/indicator/SL.UEM.1524.ZS?locations=JP.
Zhao, Poe, and Yanting. 2026. “China’s AI Labor Experiment Is Already Running.” Hello China Tech (Substack), June 23, 2026. https://hellochinatech.com/p/china-ai-labor-market.
Zou, Fangmiao. 2024. “The ‘New Trinity’ Reform of Labour Markets in Japan.” Asia Pacific Business Review 30, no. 3: 1–19. https://doi.org/10.1080/13602381.2024.2320550.



Comments