AI impacts paychecks more than payrolls, study finds
Apollo research shows artificial intelligence is slowing wage growth most in low-paying jobs.
1AI's first measurable impact
An Apollo analysis of 321 occupations found wages in AI-exposed jobs grew 6.7% slower post-2023 compared to others, with no statistically significant employment loss. Chief economist Torsten Slok notes AI's primary damage so far appears in wages rather than unemployment figures, marking a subtle but important shift in labor market dynamics.
2Wage tier disparities
The study revealed AI exposure slowed wage growth by 10.7% in the bottom 25% income bracket, with middle tiers showing 5.4% and 4.0% effects. Remarkably, the top 25% wage earners experienced no significant impact, indicating AI's influence concentrates particularly in low-skilled occupations where automation threatens routine tasks most directly.
3Novel measurement approach
Researchers matched 2015-2025 labor statistics using the Anthropic Economic Index, which tracks actual AI model interactions rather than theoretical exposure. Data limitations exist: analysis covered only 321 of roughly 800 occupations using one company's usage data. Striking findings like the 24.3% gap in service sectors come from small subsamples requiring cautious interpretation.
4Contradictory evidence
US government data showed 0.2% employment decline across 18 AI-exposed occupations amid 0.8% overall job growth. Goldman Sachs reported faster declines in AI-substitutable job postings. Former IBM HR head Diane Gherson notes companies quietly adapt by reducing hires for high-attrition, low-wage roles rather than announcing layoffs, masking the true transition.
5Accounting distortion
Gherson highlights how companies book layoffs as one-time restructuring charges (discounted by investors) while retraining appears in ongoing operational expenses. This makes workforce reductions appear financially preferable to reskilling. IKEA's success retraining call center staff as €1.3bn remote design advisors demonstrates alternative pathways.
6Economic dynamism claim
Slok argues the economy shows increased dynamism rather than contraction, with business formation at record highs. However, productivity gains beyond major tech firms remain unproven. Europe lacks comparable studies, making Apollo's wage channel findings difficult to detect in European labor data that tracks different metrics.
7Sectoral impact variations
While some service sectors showed dramatic 24.3% wage growth suppression, creative industries and high-skill professions saw minimal effects. This selective impact suggests AI transforms labor markets heterogeneously rather than uniformly. Positions heavy with routine tasks experience the sharpest wage growth deceleration.
8Long-term implications
The study indicates AI drives structural transformation rather than sudden labor market destruction. Low-skilled workers face eroding wage power as companies implement quiet strategic adaptations. Europe's measurement gap suggests global labor markets may respond differently to AI. Productivity benefits remain uncertain outside tech giants, with margins yet to show consistent improvement.
Original article: The Next Web — AI ücretleri istihdamdan daha çok etkiliyor