Important security note: Warning of attempted fraud in the name of DWS
We have detected that fraudulent individuals are misusing the "DWS" trademark and the names of DWS employees on the internet and social media. These fraudsters are operating fake websites, Facebook pages, WhatsApp groups and Mobile Apps. Please be aware that DWS does not have any Facebook Ambassador profiles or WhatsApp chats. If you receive any unexpected calls, messages, or emails claiming to be from DWS, exercise caution and do not make any payments or disclose personal information. We encourage you to report any suspicious activity to info@dws.com, including any relevant documents and the original fraudulent email. Additionally, if you believe you have been a victim of fraud, please notify your local authorities and take steps to protect yourself.
When AI makes human work more valuable
31/7/2026
Technology does not simply destroy work. Early indications are that US labor demand is shifting toward jobs where productivity—and automation—advance more slowly.
New technologies are easier to imagine than their consequences. “Very simple was my explanation, and plausible enough—as most wrong theories are!” observes H. G. Wells’s Time Traveller on encountering a dystopian future.[1] Many US companies may have recently made a similar mistake: even with artificial intelligence (AI), they still need plenty of people.[2] Technological progress does not simply reduce the amount of work. It also changes which tasks are in demand. Both economic theory and recent US data point in that direction.
Our Chart of the Week combines year-on-year changes in filled jobs and vacancies. Together, the two series point to an increase of almost one million positions in private-sector labor demand, albeit with wide differences across industries. Education and health services, professional and business services, and trade, transport and utilities led the gains. Finance, information and government recorded declines.
Separate data show employment since May 2025 rising by almost 11% in social assistance and about 5% in health care, while private education remained stable. This pattern is consistent with Baumol’s cost disease.[3] The key intuition is that, in some activities, human involvement is an essential part of the service people value. A teacher cannot keep teaching more students, nor can a care worker look after more and more people, without compromising quality. In manufacturing, by contrast, better machinery can produce more with less human labor.[4]
When industrial productivity rises, workers benefit—not only in industry itself. Higher industrial wages put other sectors under pressure. Schools, care providers and other labor-intensive services must also pay more to attract and retain staff, even though their productivity grows more slowly. If a wealthier society continues to value these services, both their relative costs and their shares of spending and employment may rise. Over time, employment can therefore grow fastest where productivity grows slowest.
“The early evidence points to task-level gains, not an economy-wide productivity miracle,” says Christian Scherrmann, DWS Chief U.S. Economist. Companies still need people to use AI effectively. That may limit widespread labor substitution, although employers cannot yet tell how far that effect will extend.[5] “More importantly, AI, like previous technologies, will change relative prices. This has the potential to shift labor toward goods and services that remain valued but are hard to automate.”
All this is illustrative, not conclusive. Many forces besides AI shape labor-market data. For now, policy and demographic trends, including aging and migration, are likely to drive health-care employment more than economy-wide Baumol effects. Even so, dystopian predictions of machines taking all the jobs increasingly look too simplistic. They are unconvincing in theory and increasingly at odds with the early evidence.
Sources: Bureau of Labor Statistics (BLS), DWS Investment GmbH; as of 7/24/26
*Year-over-year change, in thousands; seasonally adjusted. Payroll employment through June 2026; job openings through May 2026.
This information is subject to change at any time, based upon economic, market and other considerations and should not be construed as a recommendation. Past performance is not indicative of future returns. Forecasts are based on assumptions, estimates, opinions and hypothetical models that may prove to be incorrect.