🔍 Read the full analysis: Knowledge Work Adaptation To AI 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
A November 10 report from ThorstenMeyerAI.com traces the growth of knowledge and information work, from estimates that it represented nearly 29% of adjusted U.S. GNP in 1958 to its later role as a major employment category. It argues that generative AI is aimed at tasks common in office and professional jobs, but says exposure does not establish job loss and broad U.S. employment data shows no clear AI footprint so far.
ThorstenMeyerAI.com published a report November 10 tracing the growth of knowledge work and examining how generative AI could affect the office jobs built around it. The report says higher-income and clerical roles are among those with greater estimated exposure, while stressing that exposure is not the same as job loss and that overall U.S. employment data has shown no clear AI footprint so far.
The report, the eighth episode of the series Work After Work and the final episode in its first arc, sets current AI concerns against the longer shift from farm and factory work toward services and information work. It draws on economist Fritz Machlup’s 1962 estimate that knowledge-related production and distribution accounted for $136.4 billion, or “almost 29 per cent of adjusted GNP,” in the United States in 1958. That measure counted spending, not the value of knowledge itself.
The report also cites Marc Porat’s 1977 analysis of U.S. employment by the work people did, rather than the industry that employed them. Porat estimated that information workers made up 5.8% of the workforce in 1860 and 46.4% in 1970. For a more recent comparison, the report gives 2025 International Labour Organization figures: managers, professionals and technicians accounted for 62% of jobs in Singapore, 52.5% in the United States and 47.9% in Germany. Those categories are not necessarily identical to the historical definitions of information or knowledge work.
On AI, the report says estimates place clerical and higher-income jobs nearer the front of the line for generative AI exposure than many other types of work, a shift also explored in thinking about basic income as AI reshapes work. It also reports that young workers in the most exposed jobs are falling behind peers, but supplies no specific figures or study details alongside that claim. The report does not establish that AI caused those employment differences.
Infographic 01 · The number to remember
Fritz Machlup's first count of the knowledge industries, by branch.
US spending on knowledge by branch, 1958 · share of $136.4 bn
Machlup measured what knowledge cost (spending, the way national accounts count), not what it was worth. “Almost 29 per cent of adjusted GNP” is his figure for 1958. The branch values are as tabulated by Godin (2008); the labour shares are Machlup’s estimate as summarised by his publisher.
Sources: F. Machlup, The Production and Distribution of Knowledge in the United States (Princeton University Press, 1962), as tabulated by B. Godin, “The Knowledge Economy: Fritz Machlup's Construction of a Synthetic Concept” (2008); Princeton University Press, book description.
Powered by Thorsten Meyer AI
AI Reaches the Office Economy
The development matters because generative AI is associated with tasks common in knowledge work, including writing, reading, calculating, checking and communicating. If tools change how those tasks are done, the effects could reach a large share of workers in wealthy economies, not only people in technology companies. The report’s historical figures underscore the scale of the shift: work centered on information grew alongside, and in Porat’s account eventually surpassed, older employment groupings.
But the scale of exposure should not be confused with a forecast of mass unemployment. A tool may assist with, speed up or reorganize a task without eliminating a job. The report’s distinction is important for workers and employers weighing training, hiring and workplace changes: the likely effects on specific duties are not yet the same as confirmed effects on total employment.
Products Worth Considering
Canon PIXMA PRO-200S Wireless Professional Inkjet Photo Printer
Dell UltraSharp U2424H 24-Inch FHD Monitor
Brother INKvestment 6960 Wireless Color Inkjet All-in-One Printer with 11 x 17 Capabilities and 3.5 Color Touchscreen
As an affiliate, we earn on qualifying purchases.
From Knowledge Work to AI
The report traces the category of knowledge work to Peter Drucker’s 1959 book, The Landmarks of Tomorrow. Drucker described workers who earn a living through what they have learned rather than primarily through manual labor. The report presents his argument that managing people whose output depends on knowledge would require different methods from those developed for factory production.
In 1962, Machlup approached the issue by measuring the cost of producing and distributing knowledge. He grouped activity into education, research and development, communications media, information machines and information services. His estimate covered spending in 1958, so it should not be read as a direct estimate of knowledge’s economic value.
Porat later classified workers by their daily tasks, creating four categories: agriculture, industry, services and information. In his estimates, information work rose to 46.4% of the U.S. workforce by 1970. These historical measures use different definitions and methods; they provide evidence of a long-term change in work, not a direct measure of AI’s current employment impact.
““Almost 29 per cent of adjusted GNP.””
— Fritz Machlup, as quoted in the report
AI’s Job Effects Remain Unclear
The report does not identify the studies or methodology behind its estimates of which occupations have higher generative AI exposure. It also gives no specific figures, time period or comparison group for its statement that young workers in highly exposed jobs are falling behind their peers. The cause of that reported gap is not established in the material provided.
It remains unclear how quickly employers will adopt AI tools, which tasks they will automate or change, and whether any task reductions will lead to fewer jobs, different jobs or changes in productivity. The report says overall U.S. job data has no clear AI footprint so far, but does not define the data series or period used for that assessment. Its historical comparisons also rely on categories that differ across studies and years.
Watch for Employment Evidence
The report identifies no specific upcoming jobs release, research publication or policy decision tied to its claims. The next evidence to watch for is more detailed reporting on employment and hiring by occupation, especially data that distinguishes AI exposure from actual adoption and job changes. Researchers and readers will also need clear definitions of which tasks count as knowledge work and which occupations are considered exposed.
For now, the episode closes its first series arc by placing AI within a longer history of economic change. Its account supports a measured conclusion: knowledge work is a large part of modern employment, and generative AI is directed at some of its tasks, but the scale and direction of the resulting labor-market changes remain unsettled.
Key Questions
What is the news development?
ThorstenMeyerAI.com published a report on November 10, 2026, examining the history of knowledge work and the possible effects of generative AI on office and professional jobs.
Does AI exposure mean a job will be eliminated?
No. The report says exposure is not the same as job loss. AI may affect particular tasks, but the report does not establish how those changes will affect total employment.
What did Machlup’s 29% figure measure?
It was an estimate of spending on producing and distributing knowledge in the United States in 1958: $136.4 billion, or almost 29% of adjusted GNP. The report emphasizes that this measured cost, not the value of knowledge itself.
How large was information work in Porat’s estimates?
Marc Porat estimated that information workers represented 46.4% of the U.S. workforce in 1970, up from 5.8% in 1860. These are historical estimates based on Porat’s classification of workers by their tasks.
What is still unknown about AI and jobs?
The report does not provide enough detail to verify the methods behind its exposure estimates or its claim about young workers in highly exposed jobs. It also says overall U.S. job data shows no clear AI footprint so far, leaving the scale, timing and causes of employment changes uncertain.
Source: ThorstenMeyerAI.com
