AI-driven layoffs are beginning to show up in the jobs market. Goldman Sachs economists warn the fallout could last years.

Study finds 'scarring' from tech-driven displacement

Goldman Sachs economists examined past waves of technology-driven job losses and found a pattern that should unsettle workers: when machines replace tasks, the pain doesn't stop with the last paycheck. Pierfrancesco Mei and Jessica Rindels, economists and authors of the report, wrote that their analysis points to long-lived effects for displaced workers.

The impact goes beyond just short-term unemployment. The report finds what the authors call "scarring" — slower earnings growth and lower lifetime income after technology-driven displacement.

The team compared workers displaced by automation and computerization in earlier decades to people who lost jobs for other reasons. Workers displaced by technology were more likely to struggle to find comparable work, to take pay cuts, and to see their earnings climb more slowly over the following decade. Over ten years, they calculated that earnings growth for displaced workers ran roughly 10 percent slower than for peers who were laid off for other causes.

That gap matters. It means lost wages now can translate into lower raises later.

It can delay major life choices — like buying a home or even marrying — because people's finances don't recover the same way.

How bad could it get with AI?

The report's authors warned the effects may be sharper with AI than with previous technologies. "Overall, these patterns suggest that AI-driven displacement could impose lasting costs on affected workers, with substantially larger effects when job losses coincide with a recession," Mei and Rindels wrote.

Mei, who has been publicly discussing AI's labour-market impact, told colleagues that job losses tied to AI have been visible so far but remain moderate, though he cautioned the risks are skewed to the upside. Pierfrancesco Mei, economist at Goldman Sachs, noted in a separate note that faster adoption could add as much as 0.3 percentage points to the unemployment rate in 2026 on top of other effects.

Those projections show why policymakers and employers are watching adoption rates. If firms accelerate AI deployment during a weak economy, displaced workers could face a double hit: fewer openings and a glut of job-seekers with outdated firm-specific skills.

Real-world signals: hiring, markets, and surveys

There are already signs companies are adjusting headcount plans. A Harvard Business Review survey cited by market coverage found 39 percent of firms reported low to moderate reductions in staff because of AI, while 21 percent said they made large cuts and 29 percent said they're hiring less than usual in anticipation of AI's effects.

Many firms have already taken action on automation, not just talked about it. That behavior filters into hiring pipelines and entry-level opportunities, especially in white-collar fields where new graduates traditionally cut their teeth.

Investors have noticed too. Markets reacted sharply to pieces highlighting AI-driven disruption. Analyst group Citrini Research sketched a severe scenario in which agentic AI could double unemployment by 2028 and slash market value by about a third. That hypothetical sent shockwaves through tech and software stocks.

One immediate market reaction: IBM dropped 13.5 percent on a Monday following a spate of AI-related headlines — a roughly $31 billion hit in market value. Cybersecurity and other enterprise software names also swung hard as investors re-priced expectations for labor needs and software demand.

Infrastructure spending and the other side of the ledger

Not all AI effects are negative for jobs. Big capital spending on AI systems is fueling demand in some areas. David Solomon, CEO of Goldman Sachs, told TIME that AI-related investment was a nontrivial contributor to growth: in 2025, the ramp-up in AI infrastructure spending accounted for over 1 percent of U.S. GDP, he said, and the four largest hyperscalers spent up to $400 billion.

Those investments create jobs — in data centers, chip manufacturing, and cloud services — even as AI reduces the need for certain tasks inside other industries. The net balance depends on where growth occurs and whether workers who lose roles can transition into the new positions being created.

Policy choices can change the outcome

The report stresses outcomes aren't inevitable.

Policy tools could blunt or reverse scarring. The authors point to measures such as stronger severance, targeted retraining programs, placement services, taxes on automation, and worker representation in decision-making as ways to limit long-term harm.

These policy options are significant, expensive, and often face political challenges. But the report makes clear that where governments and companies choose to act will shape whether AI becomes a broad creator of wealth or an engine of durable inequality.

Frankly, the current system lacks many of these protections. And that timing is unfortunate: AI adoption is accelerating while widely available safety nets are thin.

Voices from industry

Former Medtronic CEO Bill George weighed in on labor-market shifts during media discussions, noting fewer starting positions in law firms, brokerage houses and banks. "So starting jobs in law firms and brokerage houses and investment banks are declining rapidly," Bill George said on Yahoo Finance's Opening Bid, adding that young people may pivot toward entrepreneurship because traditional entry paths are narrowing.

Meanwhile, AI firms and researchers are pointing out technical wrinkles that affect the pace of change. Anthropic highlighted that hundreds of billions of lines of COBOL remain in production across finance, airlines and government systems, suggesting technical modernization could be an avenue for AI-driven automation to cut costs but also a source of new work replacing legacy maintenance.

And some research groups paint darker turns. Citrini Research's hypothetical timeline shows a fast-feedback loop: companies fire staff after adopting AI, consumer spending drops, margins pressure firms to invest more in automation, and capabilities accelerate — a cycle that could amplify job losses if left unchecked.

What's next for workers and employers

For workers facing displacement, the immediate task is finding new income. For policymakers, the task is deciding whether to soften the fall. For employers, the question is how to balance cost cuts with social consequences and reputational risk.

There's no simple answer. Retraining can help, but the pace and scale of retraining would need to match the speed of adoption. Wage support and placement programs can reduce scarring, but they take funding and political will. Taxing automation could slow adoption or fund support; it could also push investment offshore.

Right now, the data suggests modest but real effects, with the possibility of larger shocks if adoption accelerates during a downturn. That leaves a policy window — short and uncertain — where choices matter a lot.

One-sentence paragraph for emphasis.

Bottom line: firms, workers and policymakers all face trade-offs, and the decisions made now will shape whether AI's job impact is a transitory adjustment or a generation-long handicap for affected workers.

Related Articles

"AI-driven displacement could raise the unemployment rate slightly in 2026," Pierfrancesco Mei, economist at Goldman Sachs, wrote.