Goldman Sachs thinks artificial intelligence could temporarily push about 9% of US workers out of their jobs. That estimate assumes AI lifts productivity by 15%.
In headcount terms, Goldman Sachs Research puts it at up to 15 million people over roughly a decade.
The figure is also a step up from the bank’s earlier thinking. Its previous estimate pegged displacement at 6% to 7% of the workforce.
What Goldman is actually projecting
The research is led by Goldman economist Joseph Briggs. The core idea is simple: when machines make work faster, some existing jobs disappear before new ones fill the gap.
The team built the new number from historical patterns of job disruption caused by technology. It does not rely only on net unemployment figures, which can hide a lot of churn underneath.
The US version of that churn is large. According to the research, roughly 30 million jobs are created and destroyed in the country every year.
Goldman’s historical analysis found a link between productivity and that churn. Each 1% gain in productivity from technology lifts job destruction rates by about 0.5 to 0.6 percentage points over two years.
Put a 15% productivity boost into that formula, and you arrive at the roughly 9% displacement scenario.
Who is in the blast radius
Unlike past waves of automation that hit factory floors, this one is expected to land mostly on white-collar and entry-level roles. Goldman points to technology, consulting, graphic design and administrative support as exposed fields.
Briggs said the effect is already showing up in the data. He noted AI is shaving about 10,000 to 15,000 jobs per month off payroll growth in susceptible sectors such as tech and management consulting.
The bank also sees broader macro effects. Goldman forecasts the US unemployment rate could climb by as much as 0.6 percentage points because of AI adoption.
The researchers expect that bump to ease within a few years as the labor market adjusts.
Why Goldman calls it temporary
The bank’s optimism rests on history. Around 85% of US job growth over the past 80 years came from occupations that did not previously exist, created by new technology.
Goldman expects a similar pattern with generative AI. The bank projects the technology could raise annual US labor productivity by about 1.5 percentage points once it is fully adopted, though the effects would vary widely across job categories.
What this means
For workers, the most important detail is the shape of the disruption. Entry-level white-collar roles have traditionally been the first rung on the career ladder, and Goldman’s research suggests that rung is getting wobbly.
If companies use AI to handle junior tasks, the pipeline that trains future senior staff could thin out.
The key indicators to track are the monthly payroll drag Briggs flagged and whether new AI-related occupations start appearing fast enough to absorb displaced workers. If the 10,000 to 15,000 monthly figure grows, the transition could prove rockier than Goldman’s base case.
Goldman’s upward revision from 6% to 7% to about 9% is itself a signal. The bank’s researchers now see a bigger wave of disruption than they did before, even as they continue to argue the tide will eventually turn.
Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.