AI created 640K jobs in two years but youth unemployment and white-collar displacement loom
Apr 6, 2026
Key Points
- AI created 640,000 US jobs between 2023 and 2025, but LinkedIn's head of economics cautions the figure masks temporary construction work and isn't large enough to shift overall labor market direction.
- AI-specific job postings doubled from 1.6% to 3.4% of all postings in two years, though many roles like "AI engineer" appear to be relabeled existing positions rather than genuinely new categories.
- Goldman Sachs research shows AI could automate a quarter of all US working hours with exposure concentrated in administrative, legal, and engineering roles, yet CFO surveys report essentially no negative employment effect in 2025.
Summary
AI has created 640,000 jobs between 2023 and 2025 in the US, according to LinkedIn's analysis of job posting data, yet the labor market story is more complicated than a simple net-job narrative suggests.
The headline number includes new white-collar positions like head of AI and AI engineer, but doesn't capture the temporary construction jobs tied to building data centers—a category that could dwarf permanent roles if the infrastructure buildout continues at scale. LinkedIn's head of economics, Corey Cantanga, cautions that 640,000 jobs aren't enough to meaningfully shift the overall labor market direction.
AI-specific role growth has been steep. In 2023, AI-related jobs accounted for 1.6% of all postings. By 2025, that had more than doubled to 3.4%. The title "AI engineer" in particular appears to be repurposing existing engineering hiring—companies relabeling roles to signal greenfield, flexible work rather than creating genuinely new categories. "Head of AI" positions tell a similar story: companies created 225,000 such roles in three years, though filling them requires experience that didn't exist in the AI labor market a decade ago.
The white-collar displacement risk is real but hard to measure. Goldman Sachs research suggests AI could automate tasks accounting for a quarter of all US working hours, with exposure concentrated in administrative support, legal work, and architecture and engineering. Yet a survey of 750 CFOs found AI had "essentially no negative employment effect" in 2025, and analysts acknowledge it remains difficult to separate genuine AI-driven cuts from layoffs driven by financial or operational pressures.
The broader tension: job growth in AI roles coexists with anxiety about displacement, geopolitical stress (war, oil prices pushing inflation), and the possibility that rates stay high long enough to create unemployment elsewhere. OpenAI recently released a research paper on the social contract and redistribution in the age of AGI, suggesting the industry recognizes the risk even if the near-term labor data remains mixed.