The free enterprise system’s greatest prophets saw the trade-off from the beginning. The division of labor generates unprecedented wealth while subjecting workers to menial labor that diminishes our personal and political capacities. Artificial intelligence may finally offer a way to keep the wealth while recovering what we lost.
Adam Smith’s famous 18th-century pin factory visit gave him a glimpse of the future. Workers performing fragmentary tasks could produce tens of thousands of pins a day while one worker by himself might not be able to make even a single pin. The productive benefits were too obvious to resist. If the factory owner divides the labor in his factory into its smallest components, then production would multiply beyond measure. And multiply it did. The principle Smith observed would generate more material prosperity for society over the next two centuries than in all of human history before it. But Smith also saw a cost. The worker who spends his life performing “a few simple operations … naturally loses, therefore, the habit of such [exertion of his mind], and generally becomes as stupid and ignorant as it is possible for a human creature to become.” It was clear to him that the division of labor would enrich us and diminish our human faculties at the same time.
Two and a half centuries later, we have inherited both halves of Smith’s prophecy. The Western world has now largely abandoned hyperspecialized manual labor, moving instead into service jobs: white-collar cognitive work. Yet a significant share of white-collar work feels socially pointless, often just as disconnected from meaningful ends and as stultifying as the assembly line ever was. The factory fragmented work with one’s hands; the modern office, in too many cases, has done the same to work with one’s mind.
Does AI mark the end of this era?
Artificial Intelligence promises to automate the repetitive, pin-factory office work we’d rather not do. But more importantly, it is letting ordinary people accomplish creative work in mere hours in what would have previously taken months or years. Yet a second possible corrective AI offers, less discussed than automation, may be just as transformational for the future of human work.
In his book Philosophy of Democratic Government, the French political philosopher Yves R. Simon described something that by the mid-20th century was already vanishing: the integrated, humanizing work of the family farm. For Simon, the farm exemplified a quality of labor that Smith’s pin factory had systematically destroyed.
Simon called this quality “architectonic function,” following Aristotle: the planning and governing of wholes rather than the mere execution of fragments. The farmer doesn’t execute tasks assigned by someone else. He plans the year. He reads the weather and the soil. He adjusts to nature’s resistance. He bears the consequences of his decisions. And he integrates everything he does around the needs of his household: real human goods rooted in the concrete life of his family. The work was communal as well as individual. The whole family labored together across seasons and generations, and the farmer’s daily encounter with the land kept him grounded in the physical world and its rhythms.
This kind of work, Simon argued, cultivates what is deepest in us: creativity, judgment, prudence, self-governance. He understands the entire process and exercises control over all its phases, not merely a fragment. The knowledge he gains is tacit and embodied, passed from father to son and mother to daughter, not through manuals but through shared work.
Contrast this with the assembly line. Here the wholeness shatters. The factory laborer concerns himself not with wholes but with parts, and not with ends but with motions. He is disconnected from the final product. His reasoning faculties go unused and his creativity is not merely unrewarded but actively suppressed. In Simon’s terms, he becomes a worker of parts rather than wholes, directed by distant experts and deprived of self-governance in his labor. He executes someone else’s plan without ever exercising the judgment required to form one of his own.
This pattern did not end with manufacturing. We moved from factories into offices and cubicles, positions that offer stability and comfort but rarely the experience of planning, directing, and governing a meaningful whole.
This loss extends even beyond the workplace. It is a loss for democracy itself. Thomas Jefferson believed the independent, self-directed nature of farmers made them ideal republican citizens. The farmer governs his own life and his own household and so is trained in the art of self-government. He can participate in governing his community because he already governs his farm. The hyperspecialized worker, by contrast, spends his day hardly governing anything. He himself is governed by ends set by his boss. The political scientist Vincent Ostrom, who spent his career studying self-governance, put it starkly: Democratic society depends on the “universality of artisanship” among its people. We learn to govern our work before we govern our societies alongside others. When neither work nor community offers an experience of governance, the muscles of self-rule atrophy.
AI may offer a way to rebuild them.
The common narrative treats AI as the final wave of automation: machines that will take over intellectual tasks just as earlier machines took over physical ones. On this view, AI commandeers people’s jobs and humans are left with nothing to do. But something else is happening alongside automation. AI is becoming a tool that lets people plan, direct, and execute complex projects, not by replacing their judgment but by amplifying their capacity to act on it. In a world where automation frees people from menial tasks, AI can help them pursue and achieve more ambitious projects of their own.
The people who have realized this first are the builders in the tech industry closest to the source of the new developments. What these engineers (who are no strangers to menial computer work) have discovered is that AI changes this equation. Engineers who once spent their days debugging and writing boilerplate code now find themselves freed to design and create. The clearest example of this shift is the rise of vibe coding: building software by describing what you want to an AI rather than writing code yourself. Last March, Y Combinator CEO Garry Tan stated that for 25% of the startups in his accelerator program, 95% of their lines of code are AI generated. That is not a small development for Silicon Valley’s most prestigious accelerator. But vibe coding does something more radical: It extends this creative power to people outside the industry entirely.
People with little or no programming experience are creating functional applications that did not exist before, accomplishing in hours what would have required years of learning to code. A 2025 study tested industrial engineering students with minimal web development experience on their ability to build a functional survey application using AI assistance. Of 33 participants, 24 successfully completed a working application, spending only a few hours on a task that would have previously demanded months of intensive learning.
This is democratized intelligence, available to anyone. The person selects the ends and AI executes the means. For the first time since the Industrial Revolution began, we may have instruments that could make architectonic work widely accessible again.
The problems, however, are real. AI-assisted work does not automatically restore the communal character of the family farm, where labor was shared across generations and the whole household worked together. It can encourage hyperindividuality, each person drawn deeper into projects conducted with an army of AI agents rather than alongside other people. The grounding connection to nature remains absent. The knowledge worker, even one engaged in architectonic labor, operates in a realm of pure abstraction. Regardless, as of now, the technical problems with vibe coding are not completely solved, and the apps being made often have structural issues in the code that an untrained eye that has never coded before is not able to recognize.
There is also the problem of passivity. Work can become merely a matter of passive supervision of AI rather than putting human agency at the center of the process. Furthermore, if automation proceeds far enough that significant income must come from sources like universal basic income, work may not be an attractive option at all.
Yet work is an intrinsic human good. Nobody flourishes as a couch potato or one of the human characters from WALL-E. The question is whether our economic institutions will be structured to provide the kind of work attractive enough to form an alternative to passive consumption of what AI produces for us. We need to develop enterprises people will want to spend their working lives pursuing and ways to encourage self-motivated entrepreneurship. Businesses structured to give workers genuine agency over their labor, including forms of co-ownership and distributed governance, may prove essential to preserving the architectonic character of work—even work itself—as AI transforms what that work looks like.
The opportunity is genuine. What once required years of specialized training now lies within reach of anyone willing to learn how to direct these new instruments. The deeper question is whether we will build economic institutions that encourage people to use these tools for genuine creation rather than passive consumption. Will enterprises be structured to give workers real agency? Will people be encouraged to govern something meaningful in their lives, not merely to consume and obey?
If we actively pursue these goals, we can live in a world of both meaningful work and material prosperity. People can work in ways that exercise and develop their most human faculties, preparing them for democratic citizenship in a way we desperately need.
We failed to shape the internet for human flourishing. We built instead an attention economy that exploits our weaknesses rather than cultivating our strengths. With AI, we have another chance. The tools are here. The question is whether we will build the institutions, the enterprises, and the culture that use them for what matters most: forming the kind of people capable of governing themselves and one another.
