French Treasury: AI's net effect on jobs is uncertain, but skilled professions are most exposed

The French Treasury published Trésor-Éco n°391 on June 30, 2026, arguing that the net effect of AI on employment is indeterminate due to opposing channels: substitution of workers versus increased productivity and demand. Individual gains are measured at +14% for customer service agents and +26% for software developers, but aggregate effects are unclear. Unlike past robotization which hit manual workers, generative AI affects qualified professions, with the most exposed sectors being financial and insurance activities, IT services, publishing, and professional services. The Treasury stresses that qualification level does not reliably predict impact; exposure does not mean substitution. Telemarketers, secretaries, and accountants are most substitutable, while lawyers, architects, and doctors are complementary. A concerning point is slower youth labor market entry: a study cited (Brynjolfsson et al., 2025) shows a 16% decline in employment for 22-25 year olds in exposed professions from Nov 2022 to July 2025; in French IT, employment of those aged 15-29 contracted more than gains in other age groups. Long-term, the Treasury leans toward creative destruction, but a massive substitution scenario remains possible if agentic and physical AI become widespread; more than 40% of occupations could exceed 30% automatable tasks. It recommends public policies, especially training, citing France's 18% enterprise adoption (OECD 2025) versus 20% European average, and programs like France 2030's 'Compétences et métiers d'avenir' and 'Osez l'IA' to train 15 million professionals by 2030. The note coincides with OpenAI's 'AI Jobs Transition Framework' extension in Brussels, estimating for France 13% high automation potential, 29% reorganization, 13% development, and 45% little affected. Data from Ramp and Revelio Labs show intensive AI adopters saw workforce growth of about 10% over two years, including junior positions.