History rarely moves in straight lines, but the effects of certain technologies have repeating patterns. They don’t just improve an existing process; they unlock entirely new scales of production, rearrange labor markets, and accelerate economic forces that were already gathering strength in the background. I'm going to show you three inventions, separated by more than two centuries, and see if you get the picture.

1793: Eli Whitney and the Cotton Gin

In 1793, Eli Whitney invented the cotton gin, a simple machine that separated the sticky seeds from cotton fibers. Before it, cleaning a single pound of lint by hand required hours of painstaking work. A hand-powered gin could produce roughly fifty pounds of clean cotton in a day, a leap of 50X in labor productivity.

Whitney actually hoped the invention might reduce the need for slave labor. The opposite happened. By making short-staple cotton economically viable, the gin transformed cotton into a massive cash crop. U.S. cotton production, which stood at only a few thousand bales in 1790, exploded. By 1860, the country was producing nearly four million bales and supplying the majority of the world’s cotton.

The gin moved the bottleneck. Ginning became fast; planting, tending, and hand-picking remained intensely laborious. Planters responded by expanding acreage and acquiring significantly more slaves. The U.S. slave population grew from roughly 700,000 in 1790 to nearly four million by 1860. When Congress banned the international slave trade in 1808, the value of the existing domestic enslaved population skyrocketed, and the domestic slave trade was born. A labor-saving device had, through the logic of profit, multiplied the demand for enslaved ninjas.

1913: Henry Ford and the Moving Assembly Line

One hundred twenty years later, in 1913, Henry Ford’s team installed the first moving assembly line for the Model T chassis. What had taken more than twelve hours of stationary work was reduced to roughly ninety-three minutes, a 90% reduction in assembly time.

The results cascaded like a nuclear chain reaction. Lower production costs allowed Ford to dramatically cut prices: a Model T that cost around $850 in 1908–1909 sold for $360 by 1916 and as little as $260 by the mid-1920s. Sales went from tens of thousands to well over a million vehicles a year. Automobile manufacturing ceased to be a cottage industry product for the wealthy and became a mass-market good. Employment in automotive manufacturing and its supplier industries expanded exponentially, even as the work itself became more specialized and repetitive. An industrial process that had been artisanal became a national and, soon after, a global industry.

2022: Artificial Intelligence Enters the Public Sphere

In November 2022, OpenAI released ChatGPT to the public. Built on the GPT-3.5 model and refined through reinforcement learning from human feedback, the chatbot could hold coherent conversations, write code, summarize documents, and draft text at a level that previously required skilled human effort. Within five days it reached one million users; within roughly two months it was estimated to have 100 million monthly users, the fastest growth of any consumer application to that point.

Unlike the cotton gin or the assembly line, which targeted specific physical bottlenecks, generative AI addresses a broad range of cognitive and creative tasks: writing, analysis, design, coding, research synthesis, customer service, and more. Early evidence already showed significant productivity gains in knowledge work. At the same time, the technology has triggered familiar debates about displacement, wage pressure, the concentration of returns, and the social adjustments required when a new tool multiplies the output of some workers while reducing demand for others.

Jevons Paradox

The cotton gin and the assembly line shared a structure that AI is going to repeat:

• A sharp increase in the efficiency of a previously constraining task.

• A rapid expansion of the underlying market as costs fall and volumes rise.

• A reallocation, not just a reduction, of labor. New demand is created even as certain old tasks shrink or are eliminated.

• Secondary effects that amplify existing social and economic arrangements, for better and for worse.

This dynamic is called Jevons Paradox. In the 19th century, economist William Stanley Jevons observed that more efficient steam engines did not reduce Britain’s coal consumption; they increased it. Lower cost per unit of useful work expanded the total amount of work done. The same rebound effect appears in the examples I shared above.

With the cotton gin, the efficiency gain in processing did not shrink the amount of labor required. It made cotton so much more profitable that planters planted vastly more of it and demanded more slaves to grow and pick it. Efficiency in one area produced a surge in labor demand in another part of the process.

With the moving assembly line, the dramatic cut in man-hours per car did not shrink the auto industry’s workforce. It made cars so much cheaper that demand exploded, employment in manufacturing and related industries exploded, and the total amount of industrial labor performed increased even as each individual task became more specialized.

In the 2020s the secondary effects remain to be seen. Generative AI could follow the same pattern: large gains in the efficiency of writing, coding, analysis, and design could lower the cost of mind work enough to expand the total volume of that work. More content produced, more software written, more research synthesized, more services delivered, rather than simply substituting for existing human effort. The rebound can appear as new categories of jobs, higher output from existing employees who now use AI, or entirely new markets that become viable only because the marginal cost of intelligence has fallen.

History does not supply a precise forecast. It does, however, suggest that the most consequential results of a labor-saving invention are rarely limited to the process it was designed to improve. They appear in the systems that form around the new efficiency, in the markets that expand, and in the human arrangements that prove durable or fragile under the pressure of sudden abundance. Jevons Paradox reminds us that the intuitive expectation, “this will reduce the need for X,” often fails when the efficiency gain is large enough to change what people choose to do.

The cotton gin did not end slavery; it deepened it. The assembly line did not eliminate factory work; it multiplied it and made cars ordinary. What AI will do with the cognitive and creative tasks that still define large parts of the modern economy is the open question right now. The pattern, including the possibility of a powerful rebound, is already familiar.

Now, what are you going to do about it? Moan and complain about AI, data centers and whatever? Or go out learn what the hell is going on and use it to get rich. The choice is yours.

Good luck