🔍 Read the full analysis: Where Did AI Jobs Go? 80% In Services on ThorstenMeyerAI.com
Get smart everyday buys delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
The source report says services accounted for 86% of U.S. payroll jobs outside farming in 2025, while an International Labour Organization estimate puts services at 79.6% of all U.S. workers. It links the sector’s long-term growth and rising costs to Baumol’s cost disease, then points to a study in which an AI assistant increased customer-support agents’ issues resolved per hour by 15% on average. The report does not establish that AI is eliminating service jobs.
A November 9 report from ThorstenMeyerAI.com says services made up 86% of U.S. payroll jobs outside farming in 2025, while a separate International Labour Organization estimate put the sector at 79.6% of all U.S. workers. The report connects the shift to long-running economic changes and asks whether artificial intelligence can raise productivity in work where human time and attention remain central.
The figures use different measures, so they describe related but distinct populations. The 86% figure is the author’s calculation from Bureau of Labor Statistics Current Employment Statistics annual averages; that survey excludes farm workers and the self-employed. The 79.6% figure is an ILO estimate covering the share of all U.S. workers in services. Neither means that 80% of jobs are newly created, nor that those workers have moved into services recently.
The report says service employment was already the majority of nonfarm payroll jobs in 1939, at about 62%. It puts the share at 67% in 1967 and 72% in 1979, the year U.S. manufacturing employment reached its reported peak of nearly 20 million. By 2025, the nonfarm payroll share had reached 86%. The pattern means service-sector growth preceded the manufacturing peak and continued after it.
To explain why many service costs rise even as manufactured goods become cheaper, the report draws on economist William Baumol’s theory of unbalanced growth. A live quartet still requires four musicians for a 45-minute performance. In sectors where productivity can rise sharply through machinery or scale, workers can produce more per hour; in some services, adding time-saving technology may change the service itself. The report says wages tend to rise across both types of work, putting pressure on costs in sectors where output per worker grows more slowly.
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 Human-Service Work
The issue matters because services include jobs people rely on, including health care, education, transportation, retail and customer support. If productivity improves in these fields, organizations might handle more work with the same staffing or reduce the time needed per task. That could affect service capacity and costs, but a productivity gain alone does not show that prices will fall, that workers will be displaced, or that service quality will be maintained.
The report cites a study of 5,172 customer-support agents in which an AI assistant increased issues resolved per hour by an average of 15%. That is a measured productivity result for the study’s participants, not evidence that 15% of jobs disappeared or that the same gain applies across the service economy. Whether employers use gains to serve more customers, change staffing, or alter job duties remains a separate question.
The report also uses price changes to illustrate differences between sectors. It says official, quality-adjusted U.S. television prices fell 98% from 2000, while hospital-service prices rose 274%, college tuition 188% and car repair 143%. These figures show the report’s chosen comparisons over that period; they do not, by themselves, identify the causes of each price change. The author notes that health care spending and financing differ across countries, cautioning against explaining rising costs solely through Baumol’s theory.
Products Worth Considering
Dell XPS Tower 8960 Desktop, Intel Core i7-13700, NVIDIA GeForce RTX 4060, 16GB DDR5, 512GB SSD
FLEXISPOT Electric Adjustable Workbench, 48 x 30 in, Maple
Dell 34-inch Curved USB-C Monitor S3425DW, WQHD, 120Hz
As an affiliate, we earn on qualifying purchases.
The Quartet Behind the Theory
Baumol and economist William Bowen studied the finances of performing arts organizations in their 1966 book, Performing Arts: The Economic Dilemma. Their example contrasted productivity improvements in manufacturing with a live performance that still required the musicians’ time. Baumol expanded the idea in a 1967 paper, arguing that some activities are technologically progressive while others depend more directly on labor.
As the report presents the theory, the distinction is not absolute: technology can make some services more productive, and not every factory or service job follows the same pattern. The central mechanism is that wages in slower-productivity sectors may rise alongside wages elsewhere, even if those workers cannot produce proportionally more in each hour. That can raise the relative cost of services without showing that providers are wasteful or that every price increase has the same cause.
The historical employment figures also challenge a simple account in which services grew only after factories declined. Services were already a majority of nonfarm payroll employment decades before manufacturing employment peaked in 1979. The source report’s focus is therefore both a long-term employment shift and a new question about whether AI can change productivity in some service tasks.
““The number to remember is −98%.””
— Thorsten Meyer, source report
What AI’s Job Effects Show
The cited customer-support result measures issues resolved per hour, but the source material does not provide the study’s date, the length of the trial, details of its design, or results on job numbers, wages, customer satisfaction and service quality. It therefore cannot establish whether AI is reducing employment, changing tasks, or allowing agents to handle more work without cuts.
It is also unclear how widely the reported productivity gain would apply beyond the study’s participants. Customer support is one part of a broad service sector that includes jobs with very different tasks and constraints. The source offers no economy-wide estimate of AI’s effect on service employment, and the employment shares it reports are historical measurements rather than a forecast.
The cost comparisons likewise do not isolate the causes behind changing prices. Health care spending, college tuition, car repair and television prices are affected by different market conditions and measurement methods. Baumol’s theory offers one way to understand pressures on some service costs; it does not settle why a particular bill rose or predict how AI will affect it.
Evidence Needed on Jobs
The next step is to distinguish productivity from employment outcomes. Further reporting and research would need to track whether AI-assisted service workers resolve more cases over time, whether employers change staffing or responsibilities, and whether customers receive comparable service. The source report does not identify a scheduled follow-up study or a coming employment release tied to its claims.
Readers can also watch for updated labor statistics that use clearly defined populations and time periods. Comparing payroll jobs outside farming with estimates of all workers can be useful, but the measures should not be treated as interchangeable. Until more evidence is available, the report supports a long-running shift toward services and a productivity finding in one customer-support study—not a conclusion that AI has emptied, or will empty, the service sector.
Key Questions
What does the report mean by “80% in services”?
It refers to an ILO estimate that 79.6% of all U.S. workers were in services in 2025. The report also gives a separate figure: services accounted for 86% of nonfarm payroll jobs, a measure that excludes farm workers and the self-employed.
Did services grow only after U.S. manufacturing declined?
No. The report says services were about 62% of nonfarm payroll jobs in 1939, well before manufacturing employment peaked in 1979. The service share rose both before and after that peak.
What is Baumol’s cost disease?
It is an economic theory about sectors where productivity is hard to increase because work depends directly on human time and attention. If wages rise across the economy while output per worker grows more slowly in those sectors, their costs can rise relative to goods produced more efficiently.
Does the cited AI study show that customer-support jobs are disappearing?
No. The report says an AI assistant raised issues resolved per hour by an average of 15% in a study of 5,172 agents. It does not provide evidence in the supplied material about job losses, staffing changes, wages or the study’s longer-term effects.
Do the price figures prove that service costs rise only because of Baumol’s theory?
No. The figures compare price movements from 2000, but the report does not show that Baumol’s theory caused each increase. It also points to differences in health spending across countries, indicating that other factors affect the final cost.
Source: ThorstenMeyerAI.com
