Potential Ambivalent Effects of AI on Employment, Says French Treasury

The French Treasury's Trésor-Eco note analyzes the impact of generative AI on the labor market, identifying two opposing channels: substitution of automatable jobs and complementarity that increases productivity and labor demand. Layoffs attributed to AI so far are concentrated in the most exposed sectors, such as IT and finance, and cognitive tasks are most affected, especially by large language models (LLMs). The effect varies by age: AI can reduce productivity deficits for young workers but may slow their professional insertion. Physical tasks are less exposed, while relational tasks are intermediate. Despite limited aggregate effects, AI-related layoff announcements have increased since 2025 in both the US and France, with 62% of French people viewing AI as a threat to jobs, but some announcements may be used as a pretext for structural difficulties. In 2025, AI-attributed layoffs represented 4.5% to 6.2% of announced US layoffs, and data from LinkedIn and OECD show no significant effect on hiring in exposed sectors. Historical evidence suggests new jobs created by past technological revolutions offset those lost, but public policy support, especially in training, is needed to help workers reconvert and promote AI diffusion. Investing in AI is deemed necessary for competitiveness and hence employment in a context of international competition.