🔍 Read the full analysis: AI, Agriculture, And The Changing Shape Of Work on ThorstenMeyerAI.com
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
A November 3, 2026, essay from ThorstenMeyerAI.com traces how agriculture’s share of work fell over centuries as tools, fertilizer and improved crops raised output. It uses farming as a historical comparison for AI, while presenting the idea that automation could eventually affect most work as the author’s thesis, not a forecast established by the evidence.
Thorsten Meyer published an essay on November 3 arguing that agriculture offers a historical test case for how technology can reduce the number of people needed in a sector while increasing its output. The piece traces the decline in farming employment across countries and links that history to current questions about AI and the changing shape of work; its projection that automation could eventually affect as much as 90% of a sector is Meyer’s thesis, not a confirmed outcome or independent forecast.
Meyer’s account begins with the United States, where economic historian Stanley Lebergott estimated that 73.7% of the labor force worked in agriculture in 1800. The estimate, published by the U.S. Census Bureau in a 1975 historical statistics volume, covers a labor force of 1.9 million people aged 10 and older, including about 530,000 enslaved people. Figures before 1840 are estimates rather than census counts; Meyer says “about three in four” is the more responsible summary. The measure captures a labor system that included forced labor, not just independent farm households.
The change took place over generations and through several kinds of innovation. Meyer points to the steel plough associated with John Deere in 1837, the industrial production of ammonia fertilizer after Fritz Haber’s work, and higher-yield crops, including short-straw wheat introduced in India and Pakistan in the 1960s. India’s wheat harvest rose from 12.3 million tonnes in 1964–65 to 26.4 million tonnes in 1971–72, according to the official figures cited in the essay. That is a crop-output comparison, not a measure of how many workers the technologies displaced.
The essay also distinguishes agriculture’s share of employment from the number of agricultural workers. In the United States, the number continued to rise until about 1910, reaching an estimated 11.8 million, even as the economy changed. Meyer cites an International Labour Organization estimate that 928 million people worked in agriculture worldwide in 2025, roughly one in four workers. The figures he gives vary considerably by country: about 43% of workers in India, 1.5% in the United States and 1% in Germany.
Farm Labor’s Long Decline
Agriculture matters to the AI debate because it shows that a shrinking workforce in one sector does not automatically mean that the sector produces less. Meyer cites data showing the world’s cereal harvest grew 4.2-fold from 1961, while the population grew 2.7-fold. 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 comparisons suggest that new methods can raise productivity, but they do not by themselves establish how gains were distributed or what happened to every worker who left farming.
For workers and employers, the distinction is consequential. A technology can reduce demand for some tasks while creating demand elsewhere, and the historical pattern does not guarantee that transitions will be smooth, evenly shared or repeated at the same pace. Agriculture’s shift unfolded over roughly two centuries in the United States; AI’s effects may follow a different timeline and reach jobs in many sectors at once. The essay offers a comparison to think with, not proof that the outcomes will match.
Products Worth Considering
Bose TV Speaker (2nd Gen) Compact Soundbar
Samsung Odyssey G4 25-inch FHD Gaming Monitor
Apple Watch SE 3 GPS Smartwatch, 40mm
As an affiliate, we earn on qualifying purchases.
From Hand Tools to Higher Yields
Before mechanization and modern agricultural inputs, the amount of food a household could grow depended heavily on human and animal labor, tools and weather. Where productivity was low, a large share of the population had to produce food. Historical estimates cited by Meyer show similar patterns elsewhere: Japan’s primary-sector workforce share was about 85% in 1872, while India’s 1951 census counted cultivators and agricultural laborers as about 70% of workers. These measures are not identical: some include forestry and fishing as well as farming, and some are reconstructions rather than direct census counts.
The transition came in stages rather than through one invention. Deere’s account describes a steel plough designed to shed sticky soil; later machinery mechanized tasks such as reaping and threshing. Haber filed a patent related to ammonia synthesis in 1908, and fertilizer production made a major source of plant nutrients available at industrial scale. In the 1960s, high-yield, short-straw wheat helped raise harvests in South Asia when combined with fertilizer and irrigation. Technology, infrastructure and farming practices worked together; the figures do not isolate the contribution of any one factor.
“The farm is the best test case this season has for its big question, because here a whole sector of human work has already been almost completely automated.”
— Thorsten Meyer, in the November 3, 2026, essay
What History Cannot Predict
The agricultural comparisons cannot establish how quickly AI will affect employment, which jobs will change most, or whether new work will offset losses. Meyer’s “up to 90%” statement does not specify a formal definition of a sector being automated, a method for calculating the figure, or evidence that the timeline applies across industries. It should be read as the author’s argument, not a measured forecast.
Some of the historical employment figures are estimates or reconstructions, and the source material describes differences in how countries counted workers. Output growth also does not directly reveal wages, working conditions, migration, or who benefited from higher productivity. The essay’s statistics offer a broad historical comparison, but they do not settle the likely scale or distribution of AI’s effects.
The Series Turns to Factories
Meyer presents this as the second episode of “Work After Work,” a series organized around how the economy is built. The source says the following episode will examine mechanized agriculture’s reapers, threshers and tractors as part of the story of factory and farm technology. That installment may add detail to the agricultural timeline, but the material provided does not announce a publication date.
For readers following the broader AI question, the next evidence to watch is not a single historical analogy but measured changes in tasks, employment, wages and productivity across industries. No AI-related employment outcome is confirmed by this essay; its contribution is to set out the agricultural precedent and identify the limits of using it to predict the future.
Key Questions
What is the news development?
Thorsten Meyer published an essay on November 3, 2026, using agriculture’s long decline in employment share to frame questions about AI and work.
How many people still work in agriculture worldwide?
The essay cites an International Labour Organization estimate of 928 million people in 2025, or roughly one in four workers worldwide. This is an attributed estimate, not a count reported as current for 2026.
Does the essay show that AI will eliminate most jobs?
No. The claim that automation could affect up to 90% of a sector is Meyer’s thesis. The article supplies no independent forecast establishing that outcome for AI or for the broader economy.
Did agricultural employment fall as soon as farm technology improved?
Not necessarily. The essay says the number of U.S. farm workers continued rising until about 1910, even while agriculture’s share of the workforce fell over time. A declining share and a declining headcount are different measures.
What remains unknown about AI’s effect on work?
The essay does not establish the timing or scale of AI-related job changes, which tasks will be affected, or how productivity gains and displacement might be distributed. Those outcomes remain uncertain.
Source: ThorstenMeyerAI.com
