Karl Marx believed machines would eventually turn workers into something disposable. In The Communist Manifesto, Marx and Engels wrote that industrial labor had already reduced the worker to “a mere appendage of the machine.” In Capital, Marx argued that machinery would create a permanent “industrial reserve army” of unemployed workers. As automation increased, workers would lose bargaining power while capital consolidated control. The proletariat would become poorer and more desperate.
From that condition, Marx believed revolution would follow. That prediction sits at the center of his entire framework. But history moved in the opposite direction.
For more than two centuries, automation has not destroyed the value of labor; it has multiplied it. Machines allowed a single worker to produce far more than workers in earlier eras. Instead of becoming obsolete, workers became dramatically more productive.
The result was one of the largest expansions in prosperity the world has ever seen. Living standards rose, productivity soared, and workers did not become powerless appendages of machines, they became operators of increasingly powerful tools. In other words, Marx’s prediction about automation degrading labor has already failed. Interestingly, another economist anticipated something very different.
Nearly a century before Marx wrote Capital, Adam Smith described how tools and machines expand the productive power of labor. In The Wealth of Nations, Smith explained that improvements in machinery allow a single worker to accomplish the work that once required many. His famous example was the pin factory. A small group of workers using specialized tools could produce thousands of pins in a day. Without those tools, one worker might struggle to produce a few dozen. The machine did not replace the worker. It multiplied the worker’s output.
History has largely followed Smith’s model rather than Marx’s. Industrial machinery, electricity, computers, and the internet did not create a permanently-unemployed working class. They created a far more productive one. Each wave of technology expanded what individuals were capable of producing.
But artificial intelligence introduces something even more damaging to Marx’s theory. It breaks the sequence his entire argument depends on. Marx believed automation would attack the working class first. Machines would replace manual laborers and industrial workers. As those workers lost their economic value, a large population of displaced labor would form the revolutionary force that ultimately challenged capitalism. That ordering matters, in fact, it is key.
Marx’s theory depends on the proletariat being the first group pushed aside by machines. Artificial intelligence is unfolding in the opposite direction. AI is not primarily replacing manual labor, and it is not primarily replacing low-level workers performing routine tasks. The earliest disruption is appearing higher in the hierarchy.
AI systems are increasingly capable of performing work that once belonged to executives, strategists, consultants, and organizational leadership. The kinds of roles built around synthesizing information, forming strategy, writing plans, coordinating teams, and making high-level decisions—the work of the managerial and strategic class.
This is no longer hypothetical. Researchers have already begun testing this idea. In a recent experiment described in Harvard Business Review, researchers simulated a competitive corporate environment and asked human participants and an AI system to run a virtual company. The AI made strategic decisions about pricing, product design, and market positioning using the same information available to the human participants. In many cases, the AI outperformed the humans in profitability and strategic optimization.
Meanwhile, major technology companies are actively building autonomous AI agents designed to plan, act, and execute complex business tasks with minimal supervision. At the 2025 AWS re:Invent conference, Amazon introduced a new class of “frontier agents” capable of handling complex projects for hours or even days without human intervention.
These systems are part of a broader shift toward what researchers call “agentic AI”—autonomous software systems that can perceive environments, reason about goals, and take actions on behalf of humans.
The trend is clear. AI systems are moving beyond simple automation and into roles that involve planning, analysis, and decision-making across organizations. In other words, the machine is beginning to compress the very layer of society that traditionally sits above the worker. Meanwhile, the electrician still installs wiring, the plumber still fixes pipes, the mechanic still repairs engines. Physical skill in complex environments remains difficult to automate. This is the opposite of what Marx expected.
Marx believed machines would begin by replacing muscle. What we are seeing instead is machines beginning to replace structured thought at the mid-levels of organizations. That difference matters.
Marx’s framework assumes the proletariat becomes the first casualty of technological progress. Their displacement creates the shared economic pressure that fuels class conflict. The worker becomes the center of the political story because the worker is the first to be pushed aside. But if automation begins by compressing the managerial and strategic class instead, that dynamic breaks down. The first disruption is not happening at the bottom of the hierarchy. It is happening near the middle to the top.
And this problem does not stop with Marx. Many later thinkers built similar concerns about capitalism on the assumption that automation would primarily displace labor and gradually concentrate power away from workers. John Kenneth Galbraith argued that modern industrial systems would concentrate power in large corporate structures managed by technocratic elites. But artificial intelligence is beginning to automate the very managerial functions Galbraith believed would dominate the economy.
Herbert Marcuse offered a different warning. In One-Dimensional Man, he argued that technological society would trap individuals inside massive systems of industrial control.
Even more recently, economists like Paul Krugman have raised similar concerns about the long-term effects of technology on labor. Krugman has argued that technological change can weaken worker bargaining power and contribute to rising inequality, warning that automation may increasingly substitute for human labor in large portions of the economy. Artificial intelligence may push in the opposite direction.
By dramatically lowering the cost of analysis, coordination, and production, AI allows individuals to accomplish work that once required entire organizations. A single entrepreneur—equipped with powerful AI tools—can increasingly design products, analyze markets, write software, and operate businesses with very little institutional infrastructure. Instead of trapping individuals inside large systems, AI may empower far more people to build their own.
Across these different thinkers, one assumption remained remarkably consistent. Automation was supposed to threaten workers first. Artificial intelligence complicates that belief.
The first meaningful disruption is not appearing where Marx and many of his intellectual descendants expected it. It is appearing in the very layers of analysis, coordination, and strategy that once seemed immune to mechanization.
The machine was supposed to begin by replacing the worker. Artificial intelligence is beginning by replacing the people who sit near the middle and top of organizations, deciding what the workers should do. That reversal is not just a modification of Marx’s prediction. It is the final nail in the coffin.
