🔍 Read the full analysis: Why 80% Of AI Projects Are Services on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
A November 9 analysis from ThorstenMeyerAI.com argues that services expanded in the United States well before manufacturing employment peaked in 1979. It uses Baumol’s cost disease to explain why labor-intensive services often become more expensive, and cites a study in which an AI assistant raised customer-support agents’ issues resolved per hour by 15% on average. Whether AI changes the sector’s overall employment share remains unknown.
About four in five U.S. workers were employed in services in 2025, according to International Labour Organization estimates cited in a November 9 analysis by ThorstenMeyerAI.com. The piece argues that services were growing decades before factory employment peaked in 1979 and asks whether AI can raise productivity in work where human attention is part of what customers receive.
The article gives a separate measure from the U.S. Bureau of Labor Statistics: services accounted for 62% of nonfarm payroll jobs in 1939, 67% in 1967, 72% in 1979 and 86% in 2025. Those figures are the publication’s own calculations using annual averages from the BLS Current Employment Statistics survey, which excludes farm workers and self-employed people. The ILO’s 79.6% estimate uses a broader measure of workers, so the two series describe different populations.
The analysis centers on Baumol’s cost disease, an idea developed by economist William Baumol after work with William Bowen on the performing arts. Some tasks, such as manufacturing goods, can become more productive as technology reduces the labor required per unit. Other tasks, including a live performance or individual care, require a certain amount of human time. The article’s example is a 45-minute Schubert string quartet: four musicians still contribute roughly three hours of labor.
To illustrate divergent prices, the piece reports that from 2000 to 2025 the official, quality-adjusted U.S. price of a television fell 98%, while hospital services rose 274%, college tuition 188% and car repair 143%. These are figures presented by the source, not evidence that every service became more expensive for the same reason. The article also cites a study of 5,172 customer-support agents in which an AI assistant increased issues resolved per hour by 15% on average.
Infographic 02 · Baumol's cost disease
Why the workers moved to services
Baumol's mechanism in four steps, and what it looks like in US jobs.
- 1966Baumol and Bowen: the quartetA 45-minute Schubert quartet still takes about three hours of human work
- 1967Two kinds of workWhere labour is a tool, machines cut it again and again; where labour is the product, they can’t
- WagesPay rises everywhereTeachers, nurses and musicians must earn more too, or they leave
- ResultCosts rise, people moveIf we keep buying the same mix, more and more workers end up in the slow sector
Service industries, share of US nonfarm payroll jobs
US jobs, annual average 2025 (millions)
Payroll shares exclude farm workers and the self-employed (BLS establishment survey). Health spending: national health expenditure as a share of gross domestic product.
Sources: W. J. Baumol, “Macroeconomics of Unbalanced Growth”, American Economic Review 57(3), 1967, BLS Current Employment Statistics (own calculation of shares), CMS, National Health Expenditure, Table 1.
Powered by Thorsten Meyer AI
AI Meets Labor-Intensive Services
The argument matters because services make up a large share of U.S. employment and spending, but many service tasks are difficult to scale in the same way as factory production. When productivity rises in some industries and wages rise across the economy, labor-intensive services may face higher costs even if each worker’s output changes little. That framework helps explain why the expansion of services is not simply a story of factories disappearing.
AI could alter that pattern if it lets workers handle more tasks in the same amount of time. The cited customer-support study offers a measured example of a productivity gain, but it does not establish that AI will reduce costs across hospitals, schools, repair shops or the service sector as a whole. Nor does a higher number of issues resolved per hour show by itself whether customers received better outcomes or whether fewer workers will be needed.
The distinction is relevant to workers, employers and consumers. If AI assists rather than replaces employees, it may change how work is organized while leaving demand for human judgment and attention intact. If productivity gains are substantial and widely adopted, they could affect service costs and staffing. The source frames those outcomes as open questions, not settled results.
Popular right now
KitchenAid Commercial 8-Quart Bowl-Lift Stand Mixer, White
Dyson Omni-Glide Cordless Stick Vacuum, Gold/Iron
Dyson 360 Vis Nav Robot Vacuum
As an affiliate, we earn on qualifying purchases.
The Quartet Behind the Theory
Baumol and Bowen studied the financial pressures facing theater, opera, music and dance in the 1960s. Their 1966 book, Performing Arts: The Economic Dilemma, included the example of the Schubert quartet: technological ingenuity had reduced the labor needed to make a car, but a live performance still required the musicians’ time. Baumol developed the broader economic argument in his 1967 paper, “Macroeconomics of Unbalanced Growth: The Anatomy of Urban Crisis.”
In the theory, activities differ by how readily technology can increase output per hour. In manufacturing, machinery and production methods can reduce the labor needed to produce an item. In teaching, medical care and performance, reducing the time spent with each student, patient or audience may also reduce the service being delivered. Baumol described the distinction as one of degree, not a strict division: retail, for example, gained productivity through supermarkets and self-service.
The article places the theory alongside U.S. employment history. Manufacturing employment reached almost 20 million in June 1979, according to the source’s account of the prior episode in its series, but the service share had been rising for decades before that point. The long trend continued after the manufacturing peak, which complicates explanations that treat service growth only as a consequence of factory decline.
““When someone purchases an air conditioner he neither knows nor cares how much labor went into it.””
— William Baumol, quoted from his 1967 paper
AI’s Employment Effects Remain Open
The cited customer-support result does not show whether the productivity gain will reduce prices, increase service quality, change staffing or produce different outcomes across employers. The source does not provide details here about the study’s setting, duration or how its results varied among agents, so the 15% average should not be treated as a forecast for all service work.
It is also unclear how much of the price increases cited in the article reflects Baumol’s cost disease rather than other factors. The source notes that U.S. health spending is 18.0% of the economy, compared with 12.4% in Germany, and that institutional differences can shape the bill. Its price examples establish a contrast over the stated period, but do not isolate the causes of each increase.
The employment shares also have limits: the BLS payroll figures exclude farm workers and self-employed people, while the ILO figure uses a different worker measure. The article does not claim that AI will make services a smaller share of employment, and no such outcome is established by the evidence cited.
Watch Adoption and Worker Outcomes
The next question is whether the reported gains in customer support can be reproduced in other organizations and service occupations. Further evidence would need to show not just how many tasks workers complete, but also the effects on service quality, costs, workload and staffing over time.
For the broader economic claim, readers will need comparable data on prices, productivity and employment across industries. The source’s episode presents AI’s effect on service work as a question for the season rather than a conclusion. Until more evidence is available, its study result is a specific finding from customer support, not proof of a sector-wide shift.
Key Questions
Does the source say 80% of U.S. workers are in services?
It cites an ILO estimate of 79.6% for 2025, or about four in five workers. The source also reports that services made up 86% of nonfarm payroll jobs that year in its own calculation from BLS data. These figures use different measures and should not be treated as interchangeable.
What does Baumol’s cost disease mean?
It describes how costs can rise in work that is difficult to make more productive, such as live performance or some forms of care. If wages rise across the economy while the time needed for these tasks stays similar, employers may face higher costs without a comparable increase in output per worker.
Did AI raise productivity in the cited study?
The source reports that a study of 5,172 customer-support agents found an AI assistant increased issues resolved per hour by 15% on average. That finding does not establish the effects of AI across every service industry or predict future employment levels.
Does the article prove AI will replace service workers?
No. The source presents whether AI will reduce service-sector employment or shift workers into other tasks as an open question. A productivity increase in one customer-support study does not by itself show that jobs will be eliminated.
Source: ThorstenMeyerAI.com
