🔍 Read the full analysis: The Great Move From Farms To AI-Assisted Work on ThorstenMeyerAI.com
Get the latest gadgets 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
Thorsten Meyer’s November 3 article uses agriculture’s long decline in employment share to frame a question about AI-assisted work. The figures show that farm employment fell as agricultural output grew, but the shift took centuries in the United States and the article’s forecast that AI could automate up to 90% of a sector is a thesis, not an established projection.
Thorsten Meyer’s November 3 article argues that agriculture offers a historical test case for debates about AI-assisted work: the share of Americans working on farms fell from about 74% around 1800 to roughly 1.5% today, while food production increased. The comparison shows that a sector can need far fewer workers without producing less, but does not establish how quickly AI will reshape other industries.
Meyer draws on an estimate by economic historian Stanley Lebergott, published by the U.S. Census Bureau, that 1.4 million of the 1.9 million people in the U.S. labor force in 1800 worked in agriculture. The estimate covers people aged 10 and older and includes about 530,000 enslaved people. Because figures before 1840 are estimates rather than census counts, Meyer describes the share as about three-quarters. The same series puts the 1810 share even higher, at 83.7%.
The article describes agricultural change as a sequence rather than a single breakthrough: improved equipment, industrial fertilizer and higher-yield crop varieties. One example is the steel plough made by John Deere in 1837, which the company says shed sticky soil more effectively than earlier ploughs. Another is short-straw wheat associated with Norman Borlaug’s work. India’s wheat harvest rose from 12.3 million tonnes in 1964–65 to 26.4 million tonnes in 1971–72, according to figures cited by Meyer.
Employment shares and worker totals did not move in lockstep. Meyer says the number of U.S. farm workers continued rising until about 1910, reaching 11.8 million, even as farming’s share of the workforce declined. The article also cites an International Labour Organization estimate that 928 million people worked in agriculture worldwide in 2025, about one in four workers. Agriculture’s share varies widely: about 43% in India, 1.5% in the United States and 1% in Germany, as presented in the source material.
What Farm History Can—and Cannot—Show
The central lesson for workers and employers is that higher productivity can reduce the labor needed per unit of food without eliminating the need for the sector’s output. Meyer cites a 4.2-fold increase in global cereal harvests since 1961, compared with a 2.7-fold increase in population. He also reports that U.S. farm output in 2023 was nearly three times its 1948 level, with only a small rise in total inputs. Those figures challenge the claim that fewer farmers necessarily mean less food.
For readers weighing AI’s effects, the comparison is useful but limited. Agriculture’s labor transformation unfolded over a long period and involved tools, scientific advances, infrastructure and changes in markets—not one technology acting alone. Its history cautions against assuming that automation immediately removes workers, or that workers displaced from one task will find equivalent work without disruption. It also shows why employment shares alone do not describe the whole economy: the number of workers can remain high for a time even as a sector’s relative importance falls.
Meyer’s suggestion that automation could turn a sector into one requiring up to 90% fewer human hands within a decade or a few decades is explicitly his thesis, not a measured forecast. The historical case cannot settle how AI will affect particular occupations, wages or the pace of change. Its value is as a comparison that directs attention to both productivity gains and the length and unevenness of transitions.
Products Worth Considering
Logitech MX Mechanical Wireless Illuminated Keyboard, Tactile Quiet, Graphite
Apple Watch Series 12 GPS 46mm Smartwatch
LG UltraGear 27GX704A-B 27-inch QHD OLED Gaming Monitor
As an affiliate, we earn on qualifying purchases.
From Farm Labor to Higher Yields
The article places the U.S. experience alongside other countries to show that farming once accounted for a large part of work in many economies. It cites historical estimates of about 85% in Japan in 1872, 83.5% in China in 1952 and 69.7% in India in 1951. These figures are not all directly comparable: they come from different periods, sources and definitions. Japan’s figure refers to the primary sector, including forestry and fishing, while the U.S. series measures agriculture.
In the United States, the shift took place over generations. In Germany, Meyer cites a reconstruction showing agriculture’s share falling from 43% in 1882 to 25% in 1939, while the number of farm workers remained between 8 million and 10 million. These examples underline that a shrinking workforce share does not necessarily mean an immediate drop in the absolute number of workers.
Fertilizer and crop breeding helped raise output alongside machinery. Meyer notes that Fritz Haber filed a patent in 1908 for synthesizing ammonia, a process that enabled industrial nitrogen fertilizer. In the 1960s, dwarf wheat varieties produced more grain without stalks falling over as readily, particularly when combined with fertilizer and irrigation. The article’s account is not a claim that technology alone caused every change; its examples describe several developments that accumulated over time.
“The share fell long before the headcount.”
— Thorsten Meyer, in the November 3 article
Where the AI Comparison Stops
The article does not identify a specific AI system, industry, or set of occupations that could lose 90% of human labor, nor does it provide a method for calculating that forecast. The timing is also broad: “a decade or a few decades.” It should be read as Meyer’s hypothesis, not as an outcome confirmed by the agricultural data.
Historical employment figures have limits. Some early country estimates are reconstructions rather than census counts, and national definitions differ. The 1800 U.S. figure includes enslaved people and workers aged 10 and older; Germany’s 1882 census may have undercounted women working on family farms, according to Meyer. The global 928 million figure is an ILO estimate for 2025, not a count of people whose jobs are directly comparable across countries.
The source material does not establish how AI-driven productivity gains will be distributed, whether new jobs will offset losses, or how quickly workers could move between sectors. It also does not show that the social and economic conditions surrounding agricultural change will repeat in offices or other AI-exposed work.
The Next Test Is AI Adoption
Meyer frames the farm as the opening case in a series about how economies are built; the source identifies this as Episode 2 and says Episode 3 will cover reapers, threshers and tractors. That next account may add detail on how machinery changed farm labor after the earlier developments in tools, fertilizer and seeds.
For the AI comparison, the next evidence to watch is not a single headline prediction but measured changes in task demand, employment, wages and output across specific industries. The agricultural record gives a long-run reference point, but it cannot predict those outcomes. Whether AI produces a comparable shift—and over what period—remains to be established.
Key Questions
What is the main news in the article?
Thorsten Meyer published a historical comparison linking agriculture’s decline in workforce share to questions about AI-assisted work. It is a report and argument, not a new study of AI’s effects.
How many people still work in agriculture worldwide?
The article cites an ILO estimate of 928 million people in 2025, or about one in four workers globally. The figure is an estimate, and agricultural employment shares differ substantially by country.
No. Meyer says the U.S. farm-worker total kept rising until about 1910, when it reached an estimated 11.8 million, even as agriculture’s share of the workforce fell.
Does the article show that AI will replace 90% of workers in a sector?
No. The “up to 90%” statement is Meyer’s thesis, not a confirmed finding or a forecast supported by a cited model in the source material.
Why did agricultural output rise as farm employment declined?
The article points to several accumulated changes, including machinery, industrial fertilizer and improved crop varieties. It does not attribute the productivity gains to a single invention.
Source: ThorstenMeyerAI.com
