ECONOMYJune 8, 2026

AI reshapes work, not labor’s economic role

Adam Michel

Reading time: 8 min

AI will likely boost productivity while gradually transforming jobs and labor markets over time.

The proliferation of large language models is reshaping the conversation around, if not yet the fundamentals of, economic growth, labor markets, politics and social life. At the center of that debate is a familiar concern: Will artificial intelligence shift the balance of economic power from labor to capital?

The answer to that question has enormous implications for wage growth, job availability, consumer spending and labor market stability. Both theory and evidence suggest that the answer is no, AI will not much affect the value of labor in the economy. Despite frequent claims that this time is different, AI is unlikely to fundamentally shift the balance of economic power away from human labor.

AI has a wide range of possible economic and social effects. It will likely deliver meaningful productivity gains, change the nature of work in many industries and alter the composition of jobs across the economy. These changes may unfold faster than in the past, but so far, labor markets look similar to earlier technological disruptions, suggesting adjustment will take place over decades, not years. This would follow a more familiar historical path where new tools update how we work, which increases overall demand for labor by boosting productivity.

Substitution, complementarity and demand

In theory, productivity-increasing technologies can replace or complement human labor. New technologies have always replaced some jobs, but in the process have created entirely new industries, expanded overall output and enhanced the value of human inputs.

Two questions are central here: Which effect is larger, substitution or augmentation? And will AI replace more jobs than it creates? Empirical research consistently finds that investment in new technologies is complementary, primarily augmenting, not replacing work. A recent academic review of the estimates of the relationship between capital and labor concludes that a 1 percent increase in capital-per-worker raises wages by 3 percent. Similar research shows that employment also benefits.

There are several reasons for this. Labor demand is determined by the value a worker adds to output. The more productive a worker can be, the more an employer is willing to pay. Technologies that raise worker productivity thus increase wages and employment demand.

Technological substitution in some jobs is typically offset by productivity gains elsewhere that expand output, lower prices and generate entirely new forms of labor demand. Such new markets emerge and occupations with smaller footprints grow. The substitution story itself also assumes that AI will simply eliminate lower-productivity human labor. But as economist Brian Albrecht wrote on X, large productivity differences are a normal feature of existing industries.

For example, in semiconductor manufacturing, productivity at the top firms can be 30 to 40 times that of lower-productivity firms. Yet, those lower-productivity firms and workers are not systematically driven out of the market.

Economists also find that immigrant workers are complementary to native workers, raising overall productivity and wages. If humans rarely displace other humans, there is little reason to expect new software tools to be more substitutable.

Much of the confusion also stems from how AI exposure is reported. High exposure does not mean job loss. Whether employment rises or falls in an exposed industry depends on whether those workers can use AI to become more productive. If a job is built around a few easily automated tasks, employment in that particular job type is more likely to decline. However, if AI automates some tasks in high-dimensional jobs and frees the worker to concentrate on the remaining tasks, it can lower costs and increase output. If demand for lower-cost, higher-quality output increases, firms add more workers.

Workers have not lost the historical competition with capital

In standard economic models, output is attributed to the combination of labor, capital and technology. Each component can be thought of as earning a share of national income. If, over time, capital became more important for economic output, capital’s share of national income would increase. Empirical evidence does not support this claim.

Figure 1 (below) uses data from the United States Bureau of Economic Analysis to show that the labor share of net income (net of taxes and depreciation, which better captures income actually available to workers and capital owners) is within its historical range, fluctuating above and below the average of 69 percent. Labor’s share rose gradually from the mid-20th century through the early 2000s, declined modestly thereafter and has since returned near its historical average.

In 2021, labor’s share stood at 67.1 percent and has increased ever so slightly each year since. Some readers will note that labor’s share did meaningfully decline through the early 2000s, but the leveling off at the historical trend shows that these fluctuations were not a permanent structural shift toward capital.

This stability is not consistent with a story in which technological progress steadily displaces workers in favor of capital. These trends are not just apparent in the U.S. Empirical evidence on labor shares across countries shows that technological progress has not altered the power dynamics between labor and capital globally, either. Other research also consistently finds that pay and productivity (when correctly measured) have increased at almost identical rates for many decades. This shows that workers receive a share of the productivity gains from new technologies.

A more precise interpretation of this data is that capital and labor are not primarily in competition with one another. New machines, computers or an AI model do not produce output on their own. Their use must be combined with human effort.

Lessons from past technological change

The “AI is coming for your job” framing is not new. Similar concerns have accompanied nearly every major technological advancement. From the Luddite movement’s opposition to mechanized looms during the Industrial Revolution to the anticipated job losses following the introduction of the personal computer, these fears have consistently proven overstated.

Agriculture provides the most dramatic example. In 1900, roughly four in 10 U.S. workers were employed on farms. Today, that figure is closer to 2 percent. Over the same period, agricultural output increased substantially, even as labor input declined. Farmers found other occupations without mass economic disruptions.

More extreme claims about huge disruptions to come are not supported by current evidence.

The automatic teller machine (ATM) reveals a more modern version of the same story. The number of tellers per branch fell by almost half as ATM proliferation peaked in the late 1990s and early 2000s. By reducing the cost of operating a bank branch, ATMs made it economical to open more of them. The net result was that bank teller employment actually grew. Then, mobile phone banking did what ATMs could not, cutting overall teller demand after 2010. The lesson is not that substitution never happens, but that creative destruction is hard to predict in advance and reliably generates new labor demand elsewhere, as is the case with the smartphone.

E-commerce followed the same pattern. As online retail grew in market share, it did not displace brick-and-mortar retail; the older sector simply grew more slowly. The warehousing and logistics infrastructure required to fulfill online orders more than compensated for the slower growth. According to the Bureau of Labor Statistics, between 2010 and 2025, employment for couriers and messengers and for warehousing and storage grew by 103 percent and 192 percent, respectively. Brick-and-mortar retail grew 7 percent over the same period.

A similar development can be observed in the automation of telephone switching and in advances in computing, which reduced the need for routine clerical work. The results in each of these stories are not sustained, mass unemployment, but large-scale reallocation of labor into more productive, modern industries.

Read more from economics expert Adam Michel

• Regulation and fiscal drift threaten Western prosperity
• Lowering labor taxes is essential to EU competitiveness 
• What is next after Washington exits the OECD global tax plan?

The central question is whether AI represents a fundamental break from these historical patterns. The early evidence shows that it does not.

Recent work from the Yale Budget Lab finds little relationship between AI exposure measures and changes in employment or unemployment in the first three plus years since ChatGPT’s release. Even for those expecting larger disruptions, this should not be surprising. The research finds that, historically, the effects of major technological changes unfold slowly over decades rather than in months or years.

More extreme claims about huge disruptions to come are not supported by current evidence. Nor are they consistent with the long history of technological change, in which predictions of widespread displacement have repeatedly proven overstated.

Contact us today for tailored geopolitical insights and industry-specific advisory services.

Sign up for our newsletter

Receive insights from our experts every week in your inbox.

Email*

Subscribe

Agreement GDPR

We may inform you about events and products by e-mail or other electronic means.

Information about what personal data we collect about you and how we process it (including information about how we personalize our website for you and deliver content tailored to you) can be found in our privacy policy.