SIGNAL
The early AI-labor story is a wage story, not an employment story, and the incidence is concentrated on the workers least able to absorb it.
Apollo chief economist Torsten Slok and Apollo researcher Sania Edlich split roughly 300 occupations into high- and low-AI-exposure groups and tracked them before and after ChatGPT. The result Slok shared in an Aug. 22, 2026 Bloomberg interview is the cleanest version of the data the public has: wages in AI-exposed occupations grew 6.7% more slowly than wages in low-exposure ones, and the drag fell hardest on lower-income workers.
That is the mechanism underneath the headline number. It is not layoffs. It is pay compression inside jobs that still exist, against a backdrop where the BLS sees only a 0.2% drop in a small set of AI-exposed occupations and Goldman Sachs finds openings falling fastest in highly exposed fields. The shape of the effect is bargaining power thinning out before headcount thins out.
Slok himself adds the legitimate counterweight: business formation is at a record high and the economy is more dynamic. That does not cancel the distributional point, because PYMNTS Intelligence's "Resilience Deficit" research shows Labor Economy workers, typically earning under $50,000 a year, are getting less training, less confidence, and thinner buffers as AI spreads into warehouses, restaurants, hospitality, and caregiving.
The AI-labor story is the wrong story. The right one is who can move into the augmented lane, and who is being paid less for staying where they are.
Reported by Sky for Type0, from Apollo Economist Says AI Affecting Wages More Than Employment. Read the original: pymnts.com