AI Isn't Stealing Jobs, It's Just Murdering Junior Careers, Stanford Says

Analysis of US macroeconomic data since 2022 shows unemployment in AI-exposed roles rose by 0.77 percentage points, while low-exposure sectors actually saw a 0.85 point increase. The alleged robot-pocalypse turned out to be nothing more than a general macroeconomic slump. Furthermore, enterprise adopters of automated systems expanded their total workforce by 10% over two years. Those headline-grabbing mass layoffs at tech giants were simply corporate cleanup after pandemic over-hiring and budget shuffling toward server racks.

However, the bloodbath is real for entry-level candidates. Unemployment among young graduates reached 5.6% by early 2026, with researchers like Erik Brynjolfsson noting a steep drop in demand for junior coders and customer support. Companies now route simple script writing, initial data parsing, and basic text drafting directly through neural networks, removing the training wheels that traditionally built senior engineers.

Within existing teams, productivity gains tell a skewed story. In call centers, general efficiency rose by 15%, but novice workers saw a 30% jump. Tools like GitHub Copilot accelerated coding tasks anywhere from 10% to 56% depending on system architecture. Yet researchers highlight a phenomenon called jagged capabilities: neural models handle complex algorithms brilliantly while hallucinating on elementary logic, making expert oversight from mid-level and senior staff non-negotiable.

Corporate boards are happily burning the bottom rung of the ladder to pad quarterly margins, completely ignoring who will fix the hallucinatory codebase five years from now. The industry is effectively sacrificing its future talent pipeline for cheap automated placeholders today.

Source: Stanford SIEPR