Goldman Sachs Analysts Note Signs of AI's Impact on U.S. Labor Market

In a research note dated October 22, economist Joseph Briggs reported that employment growth in sectors most affected by AI has decreased by approximately 10,000 jobs over the last three months, falling short of long-term trends. However, the shortfall has been largely balanced by increased hiring in AI-related construction roles.

The analysis highlights a statistically significant positive correlation between AI adoption and rising unemployment across service sectors, coupled with a weaker negative correlation with employment growth. While these trends suggest emerging hiring pressures, the effects have been less severe than previously anticipated earlier this year. AI-related layoffs have also seen a sharp decline since their peak in the spring.

Looking to the future, historical productivity data indicate that AI could lead to substantial shifts in the workforce. Under a scenario projecting a 15% increase in economy-wide productivity due to AI, it is estimated that around 9% of current workers could experience temporary displacement as the transition occurs. Additionally, slower hiring might present an extra employment challenge, estimated at about 3%.

Evidence from past technological advancements shows that a 1% rise in technology-driven productivity tends to increase unemployment by 0.3 to 0.4 percentage points temporarily—a phenomenon that typically dissipates over two years. Studies of 25 occupations previously affected by technology reveal an average employment decline of 25% over five years, primarily due to job losses rather than slower recruitment.

The research emphasizes that economic conditions play a critical role in determining the extent of disruption. Currently, only 23% of U.S. businesses incorporate AI in their workflows, suggesting that significant productivity impacts from AI integration are still forthcoming. A robust economy may facilitate a gradual adjustment for workers, while economic downturns could hasten automation and job losses. The baseline outlook projects a slow rise in AI-related labor disruptions, avoiding sudden shocks to employment.