Economy 6 min read By Callum Montgomery
Goldman Sachs Warns AI Could Displace One in 11 Workers, With Layoffs Likely to Hit in a Recession
Goldman Sachs economist Joseph Briggs estimates that up to 9% of workers in developed economies could be displaced by AI, but argues that the job losses will be concentrated in economic downturns rather than arriving steadily.
Up to 9% of workers in the United States and other developed economies could be displaced by artificial intelligence, according to new research from Goldman Sachs, but the bank argues that the layoffs are unlikely to arrive steadily. Instead, they are expected to cluster around economic downturns, when companies tend to restructure and cut labour costs.
In a note published on Thursday, Goldman economist Joseph Briggs wrote that «concerns of imminent AI-related labor displacement have cooled in recent months after hitting a fever pitch this spring». Under the bank's forecast that AI eventually lifts productivity by 15%, he found that up to 9% of workers could be «displaced during the AI transition», with slower hiring weighing on employment by a further 3%. That equates to roughly one in 11 workers.
The more significant finding concerns timing. Briggs examined decades of data on how technology has replaced workers and concluded that «automation-related labor disruption is limited when the overall economy is healthy but accelerates during cyclical slowdowns». He warned that «AI-related labor disruptions could well amplify the broader business cycle».
The pattern is long-established. Briggs found that the share of Americans in routine jobs — clerks, secretaries, factory machine operators and mechanics — has remained stable during expansions but declined sharply during downturns. The line drops in steps at the recessions of the early 1980s, 2001, 2008 and 2020, and does not climb back to its previous level. «Historically most of the labor headwinds from routine job automation are concentrated in economic recessions,» he wrote.
Industry-level data reinforce the point. Briggs found that «labor market displacement is more severe when the economic backdrop is less favorable». The hit to unemployment from a technology-driven productivity gain was 0.1 to 0.2 percentage points bigger when the overall jobless rate was rising. Across 25 occupations that past technology upended, employment fell an extra 4% over five years for every point the national unemployment rate climbed.
This is the second time in a year that Goldman has made a version of the argument. In October 2025, its US economists David Mericle and Pierfrancesco Mei described «jobless growth» as the likely new normal, writing that «the modest job growth alongside robust GDP growth seen recently is likely to be normal to some degree in the years ahead». They added that «history also suggests that the full consequences of AI for the labor market might not become apparent until a recession hits».
They cited a leading explanation: companies use recessions to restructure and streamline their workforces by laying off workers in less productive areas. Mericle and Mei pointed to the «jobless recovery» that followed the dot-com bust, when productivity growth remained elevated and GDP rebounded earlier than employment growth.
Briggs's estimate has risen over time. In August 2025, he and economist Sarah Dong estimated that AI could displace 6% to 7% of the US workforce if widely adopted, within a range of 3% to 14%, using the same 15% productivity assumption. By June, on Goldman's Exchanges podcast, Briggs put the figure at about 9%, or 15 million workers over a decade, describing them as workers who would be «reallocated to new positions». Thursday's note calls them displaced.
He maintains that the displacement is temporary. On the podcast he said that «even under our forecast for nine percent of workers being displaced, we'd still expect that the unemployment rate increase in any given year would be less than one percentage point».
Part of the explanation lies in what Briggs measures. Thursday's note tracks gross job losses rather than the unemployment rate, because the jobless rate hides people. «Some workers that are displaced may find jobs relatively quickly,» he wrote, «while others may decide to exit the labor force.» Historically, each 1% gain in technology-driven productivity pushed the unemployment rate up 0.3 to 0.4 points at its peak, fading to zero after two years, while raising the cumulative gross job-loss rate by 0.5 to 0.6 points. Multiplying the top of that range by 15 produces the 9% figure.
The occupation data shows how severe the impact can be within a single line of work. Briggs examined 25 jobs hit hard by past technology, including photo processors, movie projectionists, proofreaders, file clerks and bank tellers. On average, employment in those jobs fell 25% in the five years after it began to decline, «suggesting that very large job losses are possible when technology makes labor redundant». He described those numbers as «likely upper bounds» for AI, since he selected the cases where technology hit hardest. Most of the shrinkage came from people losing or leaving jobs rather than from employers hiring fewer people, with separation rates rising in 19 of the 25 occupations.
3
