The AI-and-jobs debate tends to collapse into two extremes: mass unemployment, or nothing to worry about. What’s actually happened so far is messier and more specific than either.
Our look at 2026’s layoff data found a real gap between how often companies cite AI as a reason for cuts and how often laid-off workers believe that’s actually the cause, with corporate framing running well ahead of what workers report. That gap doesn’t mean AI isn’t affecting jobs, it means the link between AI and any specific layoff is harder to prove than the headlines suggest.
The exposed jobs aren’t the ones people assume
Routine, well-defined tasks are more directly automatable than complex physical or highly relational work, which cuts against the assumption that white-collar work is inherently safer than manual labor. Entry-level writing, basic coding, and first-pass research are more exposed today than skilled trades requiring physical dexterity in unpredictable environments.
New roles exist, but they need different skills
Widely cited projections point to net positive job creation from AI overall. That’s cold comfort if your specific role is displaced and the new roles need skills you don’t have. AI oversight, workflow design, and domain expertise combined with AI fluency are the actual growth categories, a different list than what’s showing up in layoff announcements.
Roles combining real domain expertise with consequential judgment calls, and physical work in unstructured environments, remain hardest for current AI to replace, a different dividing line than white-collar versus blue-collar.
See the World Economic Forum’s Future of Jobs Report for the full data.
