The augmentation argument has a math problem.
For years, the standard counter to AI displacement concerns has been that AI creates new roles faster than it eliminates old ones. Goldman Sachs Research’s reportedly latest labor modeling doesn’t support that. According to Goldman Sachs Research, as reported by Business Insider, the firm’s model estimates approximately 25,000 U.S. positions are displaced by AI labor substitution each month, while approximately 9,000 new roles emerge through AI augmentation. The net: roughly 16,000 monthly job losses at the macro level.
That’s not a count. It’s a model. Goldman Sachs Research publishes macro-level projections built from sector data, productivity estimates, and labor substitution coefficients. The 16,000 figure is a modeled estimate, not a BLS unemployment filing or employer headcount report. It should be read alongside, not instead of, measured data sources like Challenger, Gray & Christmas’s monthly AI layoff attribution tracking, which this hub has covered across multiple cycles.
Analysis
These figures are model-based estimates from a Goldman Sachs research note, not employment count data from BLS, Challenger, or a comparable government source. Methodology differs from employer-reported layoff attribution data. Both are worth tracking; neither is definitive alone.
The earnings finding compounds the picture. Goldman Sachs research reportedly finds that AI-displaced workers accept an average real earnings reduction of approximately 3% when returning to employment, according to Business Insider’s reporting. That 3% figure is a real wage cut, adjusted for inflation, not a nominal decline. For workers already navigating a labor market where AI-adjacent roles require reskilling, the wage path back to pre-displacement earnings appears to run through a pay cut first.
The macro attribution is the most significant claim. Goldman Sachs’ modeling reportedly attributes an increase of approximately 0.16 percentage points in the national unemployment rate to AI-driven displacement over the past year. To frame that: the national unemployment rate in recent BLS reports has hovered in the 4-5% range. A 0.16 percentage point AI attribution is modest in absolute terms but represents a measurable, modeled contribution that didn’t exist in prior Goldman Sachs analyses.
A methodological note is required here. Goldman Sachs’s labor flow model estimates the pace of AI labor substitution. Challenger’s data counts employer-cited reasons for layoffs. These are different instruments measuring different things. Neither is definitive. Both are worth tracking together. This hub’s April coverage of Challenger data showed AI cited in 25% of U.S. layoffs through Q1 2026, a bottom-up employer attribution that broadly rhymes with Goldman’s top-down model without being directly comparable.
Evidence
One flag on sourcing. The Wire’s original package included “BigGo” as a co-source for this item. BigGo is a price comparison platform with no evident relevance to macroeconomic labor research. That source has been excluded from this brief entirely. All figures here trace to Goldman Sachs Research as the originating authority, with Business Insider as the reporting intermediary. If the Goldman Sachs note becomes publicly available, this brief will be updated with direct sourcing.
Watch Q2 Challenger data. If AI attribution in employer-reported layoffs holds above 20% while Goldman’s net job loss model shows continued negative flow, the pattern will have held across two consecutive quarters of independent measurement. That’s the threshold where the displacement argument moves from directional to structural.