Call center employment in the United States is 39% below its historical trend. That single number, buried in a Goldman Sachs research report, is probably the clearest signal yet that AI is moving from productivity tool to headcount reduction.
According to CNBC, Goldman analyzed labor market data across major developed economies and found that industries with high AI exposure have seen consistently slower job growth since the second half of 2022. The bank combined data from more than 800 occupations and 11 separate AI adoption surveys to map where the pressure is showing up and where it isn’t.
The answer is narrower than the headlines usually suggest. Employment in call centers, software publishing, management consulting, and advertising has dropped sharply below historical trend across developed markets. Canada’s call center sector is down 33% from trend, Germany’s down 27%. These aren’t industries where AI is just making workers more efficient. These are industries where tools capable of replacing tasks outright are already deployed and the headcount is responding.
Information and communication services, broadly defined, has also slowed across nearly all major developed economies since 2022. But Goldman notes that in most countries outside the U.S., employment in those sectors still sits near or above its long-run trend. The U.S. appears to be further along in absorbing the shock.
The entry-level finding is worth paying attention to. Goldman found that AI-related hiring pressure is strongest among workers just starting out. In Australia, a 10% increase in occupational AI exposure correlated with a 0.6 percentage point drag on annual headcount growth for entry-level roles. In the U.S., that figure was 0.2 percentage points for entry-level, compared to just 0.1 percentage points for the broader workforce. For anyone building hiring pipelines or workforce training programs, this asymmetry matters. It suggests AI is compressing the traditional entry point into knowledge work, which has long-term implications for career development and talent pipelines.
On adoption rates, Goldman’s synthesis of 11 surveys puts most major developed economies at 15% to 20% AI adoption. France, the U.S., the Netherlands, and the U.K. are leading. Italy, Japan, and New Zealand are near the bottom of developed-market adoption. Emerging markets are estimated at 10% to 15%.
So what does this actually mean for people evaluating AI’s labor market impact? The Goldman data supports a more targeted read than either the doom or the dismissal camps typically offer. AI is not uniformly disrupting all work. But in specific industries, especially those built around repetitive communication, content, and coordination tasks, the signal is already clear. And for junior workers in those fields, the window for breaking in through traditional entry-level roles is getting smaller.
