🔍 Read the full analysis: How 80% Of AI Work Is Customer-Facing on ThorstenMeyerAI.com
Get hardware and tech essentials 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
A November 9, 2026, ThorstenMeyerAI.com report says services accounted for 86% of U.S. nonfarm payroll jobs in 2025, while International Labour Organization estimates put services at 79.6% of all U.S. workers. It connects the trend to Baumol’s cost disease and cites a study in which an AI assistant increased customer-support issues resolved per hour by 15% on average. The report leaves open whether such gains will reduce demand for service workers or shift work into other tasks.
A November 9 report from ThorstenMeyerAI.com says services made up 86% of U.S. nonfarm payroll jobs in 2025, and uses that long-running shift to frame a question about artificial intelligence: can AI raise productivity in work that depends on human attention? The report cites a study of 5,172 customer-support agents in which an AI assistant increased issues resolved per hour by 15% on average, but it does not establish that the technology will reduce service-sector employment, a question explored in AI’s role in factory work.
The report draws on two different employment measures. Its calculation from U.S. Bureau of Labor Statistics Current Employment Statistics puts service-providing industries at 62% of nonfarm payroll jobs in 1939, 67% in 1967, 72% in 1979 and 86% in 2025. The survey excludes farm workers and self-employed workers. Separately, International Labour Organization estimates put services at 79.6% of all U.S. workers in 2025, or about four in five.
Its explanation centers on Baumol’s cost disease, an idea developed by economists William Baumol and William Bowen in research on the performing arts. Some work can produce more output per labor hour as technology and production methods improve, including work on the assembly line. Other work, such as live music or personal care, may require much the same human time even when the surrounding economy becomes more productive.
The report cites a study of 5,172 customer-support agents that found an AI assistant raised the number of issues resolved per hour by 15% on average. That is a productivity result in the study, not evidence that every support operation will see the same gain or that the same number of workers will be needed afterward. The report frames AI as a tool directed at service work, while leaving the effects on jobs and the quality of service unsettled; workers can also explore AI apps for working smarter.
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 the Service-Sector Cost Squeeze
The story matters because services already account for a large share of U.S. employment, while many service prices have risen faster than prices for manufactured goods. The report compares official U.S. price measures since 2000: television prices fell 98% after quality adjustment, while hospital services rose 274%, college tuition 188% and car repair 143%. These are price changes over the stated period, not a measure of what every household paid.
Baumol’s explanation is that wages can rise across the economy even when output per worker improves at different rates across industries. If teachers, nurses, performers and other service workers are to keep pace with wages elsewhere, services that cannot easily cut the time needed for each interaction may become more expensive. The theory does not rule out waste, market power or policy choices; it describes one force that can contribute to rising costs.
AI could change the productivity side of that equation for some tasks, including answering questions and handling routine support cases. But a higher number of cases resolved per hour does not by itself show whether companies will lower prices, improve service, increase output or reduce staffing. Those outcomes depend on how employers use the technology and what customers require.
Products Worth Considering
Trezor Safe 7 Crypto Hardware Wallet with Bluetooth and Color Touchscreen
Logitech MX Master 4 Ergonomic Wireless Mouse with Haptic Feedback, Pale Gray
Ledger Nano X Bluetooth Crypto Wallet
As an affiliate, we earn on qualifying purchases.
From the Quartet to Service Jobs
The report traces service-sector growth back well before the peak in U.S. factory employment. Manufacturing jobs reached nearly 20 million in June 1979, according to the report, but the service share had already risen from 62% of nonfarm payroll jobs in 1939 to 72% in 1979. The long shift into services was not simply a consequence of factories shrinking.
Baumol and Bowen’s 1966 book, Performing Arts: The Economic Dilemma, examined the finances of theatre, opera, music and dance. A year later, Baumol set out a broader theory in his paper Macroeconomics of Unbalanced Growth: The Anatomy of Urban Crisis. The report uses a 45-minute Schubert string quartet as the illustration: four musicians still contribute about three person-hours of performance, while industrial production can often be reorganized to produce more with less labor.
The distinction is not absolute. The report notes that retail changed through supermarkets and self-service, and transport made it easier for a lecturer to reach more people. Its point is that productivity gains vary by task. In customer-facing work, AI may automate or speed some parts of an interaction without removing the human time needed for complex problems, personal attention or judgment.
What the Support Study Cannot Show
The supplied report gives the customer-support study’s sample size and average productivity finding, but not its full methodology, duration, participating organizations, range of results or the precise definition of an issue resolved. The 15% average increase should not be treated as a forecast for every workplace or a direct measure of job losses.
It is also unclear how much of the gain, if sustained, would translate into lower costs, shorter waits, better customer experiences or changes in staffing. The report does not provide a long-term employment study or a breakdown of which support tasks AI handled. More broadly, its discussion of service-sector prices and Baumol’s theory does not establish that productivity differences alone explain U.S. health-care or education spending.
Evidence Needed on Jobs and Service Quality
The next useful evidence will be research that follows AI-assisted service teams over time, reports how results vary by task and workplace, and tracks staffing, wages, costs and customer outcomes alongside output per hour. That would help distinguish an initial productivity improvement from a lasting change in how service work is organized.
The report presents the employment question as unresolved: AI may take on some tasks, increase the amount of work a team can handle, or shift workers toward problems that still need human attention. No specific timeline or employment forecast is provided. Until more results are available, the customer-support finding is best read as evidence of a measured productivity gain in one study, not proof that AI will empty the service sector.
Key Questions
Does the report say 80% of U.S. workers have customer-facing jobs?
No. It says 79.6% of U.S. workers were in services in 2025, based on International Labour Organization estimates. Services include many jobs that are not directly customer-facing, and the figure does not mean that four in five workers interact with customers.
What did the customer-support AI study find?
The report says a study of 5,172 customer-support agents found that an AI assistant raised issues resolved per hour by 15% on average. The source material does not provide further details about the study’s design or results across different workplaces.
What is Baumol’s cost disease?
It is an economic explanation for why costs can rise in work that is difficult to make more productive. Wages may increase across the economy, but work such as live performance or personal care can continue to require much the same human time per service.
Does the study prove AI will reduce customer-support jobs?
No. A measured increase in issues resolved per hour does not show what employers will do with the added capacity. The report says the effect on staffing, service quality and the future mix of service work remains unclear.
Source: ThorstenMeyerAI.com
