Goldman Sachs estimates that up to 9% of workers—approximately one in 11—could be displaced during the transition to AI. Economist Joseph Briggs derives the figure from the bank’s forecast of a 15% productivity increase in the US and other developed economies. Reduced recruitment could weigh on employment by an additional 3%. Crucially, the displacement estimate captures gross job losses rather than a corresponding rise in unemployment: some workers may move rapidly into new positions, while others may withdraw from the labour force. Briggs expects the displacement to be temporary.
His analysis suggests that economic conditions will strongly influence its timing. Historically, disruption associated with automation has been relatively contained during expansions but intensified in recessions, as companies reorganised and reduced labour costs. The share of Americans employed in routine occupations fell sharply during downturns in the early 1980s, 2001, 2008 and 2020, without returning to its previous level. Across 25 occupations severely affected by earlier technologies, employment declined by an average of 25% over five years after beginning to fall. Briggs regards these deliberately selected, hard-hit cases as likely upper bounds for AI.
The immediate impact remains smaller than feared earlier in 2026. Only 23% of US workplaces have integrated AI into routine operations, leaving most anticipated productivity gains ahead. Economic and labour-market resilience has also reduced the incentive to pursue aggressive staffing cuts. A downturn, Briggs warns, could accelerate AI-driven displacement and thereby deepen the broader economic slowdown.
