The figures are in, and they are shaking up both sides of the debate. The World Economic Forum's 'Future of Jobs 2026' report, published on March 28, confirms the trend identified in 2020: artificial intelligence and automation will eliminate 85 million jobs by 2028, but will create 97 million, a net positive balance of 12 million jobs. The problem, therefore, is not the overall volume but the distribution: the jobs destroyed and the jobs created are not in the same sectors, nor in the same countries, nor at the same qualification levels.
High-growth professions reveal an unexpected map. The 'prompt engineer' role — a specialist in crafting instructions for AI models — has seen its job postings multiplied by 14 in two years on LinkedIn. 'AI trainers,' responsible for supervising and correcting models, represent 340,000 available positions worldwide. However, the most sought-after profiles are not all technological: 'ethical AI auditors,' 'human-AI interaction designers,' and therapists specializing in automation-related anxiety are among the top ten fastest-growing professions.
Remote work, accelerated by the pandemic, is getting a second wind thanks to AI. According to a Stanford study published in February 2026, 42% of full-time jobs in OECD countries are now performed in hybrid or fully remote mode, up from 27% in 2023. AI-powered collaboration tools — real-time translation, automatic meeting summaries, predictive project management — have eliminated most of the friction that hampered distributed work. The geographical consequence is significant: Silicon Valley companies are now recruiting massively in Poland, Vietnam, and Kenya, where salaries remain three to five times lower for comparable skills.
Professional retraining has become the number one political issue in several countries. France has doubled the budget of France Travail (formerly Pôle emploi) in 2026 to fund 500,000 accelerated digital skills training programs. Germany has launched the 'Qualifizierung 4.0' program with a budget of 4.2 billion euros. In the United States, the 'AI Workforce Transition Act' provides tax credits of $10,000 per retrained employee. Despite these efforts, the time lag remains cruel: training an accountant in machine learning takes eighteen months; their job can be automated in three weeks.
The paradox of 2026 can be summarized as follows: there have never been so many unfilled job openings (35 million in the OECD) and so many workers in transition (48 million). The problem is not a lack of work but a mismatch between supply and demand. Educational systems, designed to train uniform cohorts over long cycles, struggle to keep pace with an economy where key skills are renewed every three years. The solution will lie in continuous training, micro-certifications, and a culture of lifelong learning — or it will not be found.
