Growth without work: AI is restructuring jobs faster than safety nets can see it

Rest of World investigated the human cost of the AI labor transition across multiple countries. The finding: displacement is happening invisibly—inside firms, through attrition and role compression, not mass layoffs—and policy is blind to it.

In plain terms: AI is changing jobs faster than governments are noticing. Workers are losing roles inside companies before the statistics catch up - and the safety net isn't designed to catch them.

The fact

Rest of World investigated the human cost of AI-driven labor restructuring across multiple geographies. The core finding: displacement is occurring through role elimination, task compression, and attrition within firms rather than headline layoff announcements. This makes it nearly invisible to standard unemployment metrics even as its effect on individual workers is immediate and severe.

Our take

This is the policy blind spot the ILO and others have been warning about, now documented at street level. Aggregate employment numbers stay stable because AI-driven displacement happens through attrition and reclassification—not the mass-layoff events that generate political response. The workers most exposed—entry-level, repetitive, data-entry, and tier-one support roles—don’t produce the kind of unemployment spike that triggers rulemaking. The ILO estimates 80 million workers across ASEAN alone are at exposure risk; Rest of World shows what that exposure actually looks like for individuals. The gap between what policy can see and what is actually happening to workers is widening.

What to watch

Whether EU AI Act labor impact assessment obligations—or equivalent US rulemaking—generate enough mandatory employer reporting to make this invisible displacement statistically visible.