AI’s Employment Toll Is Emerging, but the Evidence Remains Mixed

There are growing signs that artificial intelligence is changing employment, hiring, and the path back to work after a job loss. But the evidence does not yet support a simple claim that AI is damaging the entire job market.

Empirical proof linking AI automation to broad job market decline remains hard to establish. New data points to real effects in certain industries and for certain workers, while other research finds no economy-wide displacement. The clearest picture so far is uneven: AI may be helping productivity in some sectors while making employment transitions harder for people whose jobs are exposed to the technology.

Productivity gains are appearing alongside weaker employment

US-based financial researcher Eric Basmajian wrote that “AI productivity gains are showing up” in jobs data, though only about 15 percent of jobs seem to be impacted so far. He pointed to information, professional services, and finance as the three heaviest AI adopters, saying their productivity trend rose from 2.2 percent to 5.0 percent since 2022.

That productivity increase coincided with a high AI adoption rate and incremental gains in workplace productivity. But the numbers come with an important warning: productivity measures do not always reflect better work, stronger quality, or higher revenue. In some cases, the figures may reflect lower headcount instead.

Another analysis showed a rising connection between an industry’s AI exposure and unemployment in non-tech professional sectors, including finance, telecommunications, technical services, and administration. The data covers the United States, and the same relationship does not show up elsewhere. The link visible in the US between AI use, productivity growth, and falling employment is absent in other places.

That makes the finding important, but narrow. It points to changes in specific industries rather than proof that AI has already caused widespread unemployment across the economy.

Workers displaced by AI face a longer road back

Goldman Sachs economist Pierfrancesco Mei found that workers displaced from AI-exposed jobs since 2025 took about a month longer to find new jobs than other displaced workers. They were also roughly 10 percentage points more likely to remain unemployed a year after losing their jobs.

Those effects match the size of the damage Goldman found during earlier waves of technological change. Mei called the effects of those earlier waves “long-lasting,” and Goldman has tracked four decades of Americans who lost jobs to technology.

The impact also differed by age. AI-displaced workers age 35 and older took about five weeks longer than other displaced workers to return to full-time work. For workers under 35, the difference was about three weeks.

The Labor Department survey used in the analysis counted 7.4 million workers displaced from 2023 through 2025. The Bureau of Labor Statistics reported in August that the prior survey covered 2021 through 2023. Among people who had held their jobs for at least three years, 72.9% of workers ages 25 to 54 were working again by January 2026, compared with 57.3% of those ages 55 to 64 and 38.6% of those 65 and older.

Workers who found full-time jobs after losing long-term positions also saw weaker pay outcomes. About 49% were earning as much as or more than before, down from about 62% in the prior survey. Goldman’s earlier research found that technology-displaced workers experienced slower real earnings growth and higher odds of another period without work during the 10 years after losing their jobs.

Young workers may face the biggest hiring squeeze

Stanford economists Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen used ADP payroll data to examine workers ages 22 to 25 in occupations most exposed to AI. Employment for that group was about 19% below where it would have been if it had kept pace with similarly aged workers in less-exposed jobs.

“Experienced workers show no comparable gap,” the researchers wrote. They found that the adjustment appears to operate through reduced hiring of young workers rather than increased separations. That distinction matters: companies may be cutting off entry points instead of dismissing large numbers of experienced employees.

Goldman found that younger, college-educated, and urban workers suffered cumulative real earnings losses about half as large as other technology-displaced workers. Workers with less time in the job they lost also did better. Mei wrote that younger workers “may be able to adapt more flexibly to AI disruption,” which Goldman called an encouraging sign amid concerns about recent graduates.

Still, Goldman cautioned that it is “still too early to assess the impact along all of these dimensions,” including education and geography. Its data covers people who already had a job to lose, and no survey question asks whether AI caused someone to lose that job. Goldman instead built a score using an occupation’s AI exposure, an industry’s AI adoption rate, and whether the position was eliminated, treating the top 30% as likely AI-displaced.

The case against a full AI job-loss crisis

Several findings push back against the strongest claims about AI and employment. Goldman found no meaningful gap between the two groups for job losses before 2023, the year AI adoption began to rise. The Stanford researchers wrote that they “do not see widespread, economy-wide job displacement associated with AI.”

The Yale Budget Lab also found that measures of AI exposure, automation, and augmentation showed no sign of being related to changes in employment or unemployment. These findings do not erase the problems facing workers in AI-exposed jobs, but they show why the wider labor market picture remains unsettled.

Employers cited AI in 120,136 announced job cuts through September, about 21% of all cuts and the top reason given. That figure shows how often companies connect layoffs to AI, but it does not prove that every affected job disappeared because of automation.

Efficiency can carry a hidden cost

The debate also reaches beyond layoffs. Companies that use AI to eliminate variation and maximize efficiency risk weakening the human experimentation and creativity that support innovation.

Corporate America has run the experiment of pursuing perfect uniformity before, and the results were poor. William Abernathy of Harvard Business School called the outcome the productivity dilemma: increased efficiency came with decreased innovation.

Process management programs such as Six Sigma reduce variation in company routines. Management scholars Mary Benner and Michael Tushman examined more than two decades of patents and quality certifications in the paint and photography industries and found that the more process management a company used, the more its innovation relied on what it already knew instead of creating new approaches.

That history offers a warning for AI adoption. A company can gain productivity while losing experimentation, new ideas, and opportunities for younger workers. The employment effects may be concentrated today, but the choices businesses make about efficiency will shape whether AI expands human work or narrows it.

Stay connected via Google News

Follow us for the latest AI News