TL;DR
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
A November 9, 2026, article from Thorsten Meyer AI examines why services account for about four in five U.S. workers, linking the shift to decades of economic change and William Baumol’s theory of productivity differences. It cites a study in which an AI assistant increased customer-support agents’ issues resolved per hour by 15% on average, while leaving open how widely such gains will spread and what they mean for jobs.
Services accounted for about 79.6% of U.S. workers in 2025, according to International Labour Organization estimates cited in a November 9 analysis by Thorsten Meyer AI. The article argues that the shift toward service work began well before U.S. factory employment peaked in 1979, and points to a study where an AI assistant raised customer-support agents’ issues resolved per hour by 15% on average.
The article’s central explanation is Baumol’s cost disease: productivity can rise rapidly in industries where machines and new processes reduce the labor required to make a product, while some services still need roughly the same human time to deliver. A live string quartet, for example, still requires four musicians to perform a piece that takes about 45 minutes. The article uses that comparison to explain why wages and costs can rise in labor-intensive fields even when their output per worker is difficult to increase.
Its employment figures show that services did not grow simply because factories contracted. U.S. Bureau of Labor Statistics payroll data, as calculated by the article, put service-providing industries at 62% of nonfarm payroll jobs in 1939, 67% in 1967, 72% in 1979 and 86% in 2025. That payroll measure excludes farm workers and self-employed people; the broader ILO estimate for all workers puts the service share at 79.6% in 2025.
The article also compares prices from 2000 to 2025, citing official, quality-adjusted U.S. figures: television prices fell 98%, while hospital services rose 274%, college tuition 188% and car repair 143%. These figures are presented as illustrations of different productivity patterns, not as proof that every price change has a single cause.
Why 80% of AI Work
Ended Up in Services
Services became the dominant source of U.S. work through decades of economic change. AI may boost productivity in some service jobs, but its wider effects on costs and employment remain uncertain.
A long shift, not a sudden switch
Services were already most nonfarm payroll jobs before manufacturing employment peaked.
Why service costs can rise
Baumol’s cost disease describes how productivity advances at different speeds across industries.
- Machines and new processes can reduce labor needed to make products.
- Television prices fell 98% from 2000 to 2025.
- A live quartet still needs four musicians for a 45-minute performance.
- Hospital services rose 274% and college tuition 188% over the same period.
What the AI result shows—and doesn’t
One customer-support study offers a measured example, not a forecast for every service occupation.
An AI assistant helped 5,172 customer-support agents resolve 15% more issues per hour on average.
The result does not establish better service quality, lower prices, or comparable gains in health care and education.
Businesses could serve more people, respond faster, change staffing, or pursue lower costs.
AI Meets Labor-Intensive Services
The issue matters because most U.S. workers are in services, and many public concerns about household costs center on services such as health care, education and repairs. If technology can make some service work more productive without reducing the quality people value, it could affect how much work is needed per customer and how quickly providers can respond. The customer-support result cited in the article offers one measured example, but it does not establish that similar gains will occur across hospitals, schools or other fields.
The economic stakes extend beyond prices. Higher output per hour might allow a business to handle more requests with the same staff, reduce waiting times, or change staffing needs. Which outcome follows depends on how employers deploy the tools, whether customers demand more service as it becomes easier to provide, and whether workers’ roles change. The article frames this as an open question, rather than evidence that AI will eliminate service-sector employment.
Price comparisons also caution against treating technology as the only explanation for expensive services. Haircuts and car repairs rose substantially in the cited period, while the article notes that U.S. health spending was 18.0% of the economy, compared with 12.4% in Germany. Those spending shares do not by themselves explain the difference, but they suggest that institutions and national choices can shape costs alongside productivity pressures.
Products Worth Considering
Samsung 34-inch ViewFinity S50GC Ultra-WQHD Monitor
Apple iMac (2024) 24-inch All-in-One Desktop with M4 Chip, 16GB Unified Memory, 256GB SSD, Green
Dell UltraSharp U2424H 24-Inch FHD Monitor
As an affiliate, we earn on qualifying purchases.
From Orchestras to the Wider Economy
Economists William Baumol and William Bowen examined the finances of performing arts in their 1966 book, Performing Arts: The Economic Dilemma. Their example was that a 45-minute Schubert quartet required about three musician-hours, a workload that had not been reduced in the way factory labor could be. In 1967, Baumol developed the idea in an American Economic Review paper, describing an economy with activities where productivity advances quickly and others where labor remains central to the service.
The distinction is not absolute. The source notes that retail changed through supermarkets and self-service, while technology can also alter how services are delivered. The argument is about differences in the scope for productivity growth, not a claim that services never improve. Where one sector can produce more with fewer labor hours, while another cannot do so as readily, wages may still rise across the economy and put pressure on the latter’s costs.
The employment timeline is also relevant to the interpretation. U.S. manufacturing employment reached almost 20 million in June 1979, according to the source article, but service employment had already formed a majority of nonfarm payroll jobs decades earlier. The source therefore challenges a simple account in which service work expanded only after manufacturing declined.
““Human ingenuity has devised ways to reduce the labor necessary to produce an automobile, but no one has yet succeeded in decreasing the human effort expended at a live performance of a 45-minute Schubert quartet much below a total of three man-hours.””
— William Baumol, quoted in Thorsten Meyer AI’s account of his 1967 paper
How Far AI Gains Will Reach
The cited 15% customer-support productivity gain applies to a study of 5,172 agents, according to the source, but the supplied material does not name the study, give its publication date, or explain its methods, duration or participating organizations. It is not clear from the information provided how representative the participants were or whether the result persists over time.
There is also no evidence in the supplied material showing that AI has already reduced total service employment, lowered consumer prices, or produced comparable productivity gains in fields such as health care and education. A higher number of issues resolved per hour is one measure of output; it does not, on its own, establish the quality of the answers or the effect on workers’ pay and employment.
The price and employment statistics have different scopes and comparison bases. The BLS payroll shares cover nonfarm payroll jobs and exclude self-employed workers; the ILO share cited covers all workers. The article’s price figures compare 2000 with 2025, but the material does not detail every index method or explain the separate causes behind each price increase. The figures describe trends, not a complete account of what drove them.
Evidence Needed on AI’s Effects
The next step is to see whether the customer-support finding is replicated and whether similar tools produce measurable gains in other service occupations. Useful evidence would distinguish productivity from job cuts, track changes over time, and report service quality alongside the number of tasks completed. The source article does not identify a scheduled follow-up study or a specific policy announcement.
For readers following the broader argument, future employment and price data can test whether the long-running movement toward services continues and whether particular services become more productive. The key question is not only whether AI can complete more tasks per hour, but how organizations use that capacity: to serve more people, change staffing, improve response times or reduce costs. The material available does not yet settle which outcome will predominate.
Key Questions
What does the claim that about 80% of U.S. workers are in services mean?
The source cites an ILO estimate of 79.6% for 2025, covering all workers. It separately calculates that service-providing industries made up 86% of U.S. nonfarm payroll jobs that year; that payroll measure excludes farm workers and the self-employed.
What is Baumol’s cost disease?
It describes how costs can rise in work where productivity is difficult to increase, even as wages rise across the economy. A live performance is a common example: reducing the time musicians spend performing can change the service itself.
Did AI increase productivity in every service job?
No. The source cites one customer-support study involving 5,172 agents, where an AI assistant increased issues resolved per hour by 15% on average. The material does not establish that the same result applies to other occupations.
Does the article show that AI is eliminating service jobs?
No. It raises the possibility that AI may change service work but does not provide evidence of an overall decline in service employment caused by AI. How businesses use productivity gains remains uncertain.
Source: Thorsten Meyer AI
