Goldman Sachs puts a number on it: AI is cutting 16,000 jobs a month

AI is a net negative for U.S. employment at ~16,000 jobs lost per month, per new Goldman Sachs research: AI substitutes ~25,000 roles monthly while creating or augmenting ~9,000 — the first major bank to publish a net monthly figure.
Entry-level workers under 30 are overrepresented in routine white-collar tasks — data entry, customer service, legal support, billing — most susceptible to substitution; a 1-standard-deviation increase in AI substitution exposure widens the entry-level wage gap vs. experienced workers by ~3.3 percentage points.
Despite the sector-level churn, the U.S. economy added 178,000 jobs in March and unemployment fell to 4.3% — macro data that continues to diverge from the concentrated displacement signals in tech and white-collar sectors.

A separate Goldman Sachs research note tracking tech-displaced workers since 1980 finds real earnings remain 10 percentage points below non-displaced peers 10 years after job loss, with effects worsening significantly during recessions.
Cascading consequences include slower wealth accumulation, delayed homeownership, and lower likelihood of household formation; workers who retrained into higher-"abstract content" roles fared substantially better.
Economists at NYT who previously dismissed AI's labor market threat now call disruption likely — and warn policymakers are unprepared, urging urgent modernization of unemployment insurance and worker transition programs.

Anthropic's Head of Economics Peter McCrory finds that while coding tasks are 94% theoretically automatable by AI, observed AI usage in coding roles covers only ~30% of tasks today — a structural gap between capability and deployment.
A new BCG report classifies U.S. jobs into six disruption tiers: only ~12% face direct substitution risk; ~34% have limited exposure; ~23% will treat AI as a baseline embedded tool — BCG's headline: AI will reshape 50–55% of jobs but eliminate only 10–15% within 4–5 years.
A Yale/NBER paper argues AGI will automate "bottleneck work" essential to growth (energy, infrastructure, science) but largely bypass "supplementary work" — arts, hospitality, customer support — not because AI can't do it, but because the compute cost isn't justified; the paper predicts labor's share of GDP converges toward zero as compute becomes the scarce input regardless.

A new Brookings report warns AI is eroding not just individual jobs but the career pathways that allow 15 million workers without four-year degrees to move into higher-wage roles; nearly half of pathways between gateway and destination occupations are highly AI-exposed.
Geographic hotspots with highest gateway-pathway AI exposure include Palm Bay, FL (35.5%), Cape Coral, FL (34.7%), and Albany, NY (32.8%); Midwest metros like Cincinnati and Milwaukee show lower exposure.
Middle management postings are down ~42% and entry-level postings down ~30% since 2022, with Unilever, Oracle, Nestlé, and Meta among firms actively cutting or freezing; defense, cybersecurity, AI/ML, and healthcare remain active hiring sectors.
The Federal Reserve finds 41% of the U.S. workforce now uses generative AI for work — up 31% year-over-year — with financial services (63%) and professional services (62%) leading; only 12% use AI daily, suggesting adoption is still in early scaling phases.

AI-driven layoffs are largely concentrated in the U.S.: China's national employment mandate (~5.5% urban jobless rate target), lower labor costs (algorithm engineers earn ~$35K/yr vs. ~$300K at U.S. peers), and lower enterprise digitalization have insulated Chinese workers.
Alibaba is an exception — headcount fell more than 30% attributed to AI restructuring; Tencent and Huawei modestly increased headcount.
Chinese engineers at U.S. firms face compounding pressure: AI displacement risk combined with visa instability if laid off is pushing many to return to China rather than seek re-employment in the U.S. market.

Get Is AI replacing humans on job markets? in your inbox