The double shock of AI: youth employment in the most affected professions fell by 16%

Summary

Researcher Axelle Arquié argues that generative AI has disproportionately affected young workers aged 22-25 in the most AI-exposed occupations, resulting in a 16% relative decline in employment. Arquié notes that the automation of high-value cognitive tasks is reshaping the capital-labor balance and creating an imminent fiscal 'scissors effect' in public budgets.

Why it matters

If the 16% relative drop in youth employment in AI-exposed roles is real and accelerating, it signals that this is no longer hypothetical but a measurable trend in labor markets. The article’s focus on the simultaneous rise in public service demand (such as retraining and social safety nets) alongside a decrease in taxable jobs (revenue loss) elevates the fiscal policy debate beyond the conventional 'employment vs. productivity' framework.

Highlights

• Since the widespread adoption of generative AI, there has been a 16% relative decline in employment among young workers aged 22-25 in the most AI-exposed occupations (according to the article’s headline claim).
• Current large language models (LLMs) are still characterized as 'narrow intelligences.' These models lack causal reasoning capabilities, produce hallucinations, and struggle with out-of-distribution problems. The author notes that these deficiencies serve as concrete brakes on full automation.
• Agentic AI (multi-LLM architectures equipped with RAG and external tools) extends automation beyond text to visual, audio, and multi-step workflows. However, inference costs and a lack of explainability still limit corporate adoption in critical tasks.

Critical take

The 16% figure is claimed in the article’s introduction without providing methodology, sample size, or a counterfactual scenario. This prevents readers from assessing whether this decline is due to AI-driven displacement or general macroeconomic weakness in the 2025-26 period. Furthermore, the article cuts off before presenting the analysis of the fiscal 'scissors effect' or a concrete design for an automation tax, indicating that the text’s policy-focused core is insufficiently developed.