Chapter 19: The Automation Question

In 1930, John Maynard Keynes made a prediction. Within a century, he wrote, technological progress would solve the "economic problem"—humanity's age-old struggle against scarcity. Our grandchildren would work perhaps fifteen hours a week, devoting the rest of their time to leisure, culture, and the art of living well.

We are now past Keynes's deadline. Productivity has increased beyond what he imagined. We could, in principle, produce enough for everyone with far less labor than we currently employ. Yet we work as many hours as ever—more, in many cases. The economic problem seems unsolved; the art of living well, indefinitely deferred.

What went wrong? And what happens if automation finally delivers on Keynes's promise—not gradually, not manageably, but abruptly, through artificial intelligence that can perform most cognitive labor?

The automation question is not new. Every generation has feared that machines would steal jobs. Every generation, so far, has been proven wrong—or at least premature. But this time might be different. And understanding why requires examining both the history of automation anxiety and the genuinely novel elements of the current moment.

The History of Being Wrong

The Luddites of early nineteenth-century England smashed textile machinery, believing it would destroy their livelihoods. They were right about their own jobs—hand weavers did become obsolete—but wrong about employment in general. The factories that displaced craftsmen employed far more workers than the cottages they replaced.

Every subsequent wave of automation triggered similar fears. Tractors would devastate agricultural employment (they did, for farmers; but workers moved to factories). Assembly lines would make human workers redundant (they shifted work, not eliminated it). Computers would automate offices into emptiness (yet service employment exploded).

The pattern seemed clear: technology destroys specific jobs while creating more jobs overall. Workers displaced from one sector flow to another. The economy adjusts. Unemployment is temporary, transitional, manageable.

Economists developed a framework to explain this resilience. Automation reduces costs, which reduces prices, which increases demand, which creates jobs in adjacent sectors. Workers freed from routine tasks move to more complex, creative, interpersonal work. The economy climbs an endless ladder of sophistication, always finding new work for human hands and minds.

This framework has been remarkably robust. For two centuries, employment adapted to technological change. Prophets of technological unemployment were consistently wrong.

Until now?

Why This Time Might Be Different

The argument that "this time is different" must overcome a powerful historical prior. But several factors distinguish the current moment.

Breadth of capability: Previous automation technologies were narrow. A loom could weave cloth but not pick cotton, drive a wagon, or sell the finished product. Automation proceeded sector by sector, allowing displaced workers time to transition. Large Language Models and general-purpose AI systems are broad by nature, not narrow by function. They can write, analyze, code, advise, create. The number of tasks potentially automatable is vastly larger than in previous waves.

Speed of deployment: Industrial automation required building factories, installing machinery, training operators. It proceeded over decades. Software deployment is nearly instantaneous. An AI system that works is available everywhere, immediately. The adjustment time that cushioned previous transitions may not be available.

Cognitive labor: Previous automation primarily affected physical labor—farming, manufacturing, routine tasks. The jobs that remained, and that grew, were cognitive: analysis, communication, creativity, care. But AI targets cognitive labor directly. If machines can write, analyze, and create, the refuge that absorbed previous displaced workers may not exist.

Consider the concrete examples emerging now. Law firms use AI to review contracts that once employed armies of junior associates. Newsrooms deploy AI to generate routine sports and financial coverage. Customer service increasingly routes through chatbots before reaching humans. Graphic designers compete with image generators. Programmers find their code auto-completed, then auto-written. Each individual case seems manageable; the aggregate pattern is striking.

Capital-labor substitution: The economic ladder worked because new technologies, while displacing some workers, created demand for other workers to build, maintain, and complement the machines. But AI systems are themselves largely software—produced by a small number of highly skilled workers, replicated at near-zero marginal cost. The labor share of production may decline in ways previous automation did not cause.

None of this is certain. Technological forecasting is notoriously unreliable. New categories of work may emerge that we cannot currently imagine—just as "social media manager" was unimaginable in 1990. But the arguments for exceptionalism are serious, and prudence demands engaging with them.

The Labor-Value Nexus

Here is the deeper issue: our entire economic system is built around the assumption that most adults will earn their living through labor.

The wage is not merely an economic transaction; it is the primary mechanism by which participation in the economy is earned. You work, you receive income, you spend income, you participate. Those who cannot or do not work—children, retirees, the disabled—are supported through transfers that depend, ultimately, on workers' production.

This system creates what we might call the labor-value nexus: the idea that income is earned, that earnings reflect contribution, that contribution is measured in work. This nexus provides not just economic coordination but moral meaning. Work is not only how we get money; it is how we earn our place in society, how we structure our days, how we define our identities.

Feminist economists have long argued that this nexus was always broken. The ten to thirteen trillion dollars of unpaid care work performed globally each year—disproportionately by women—is essential to economic functioning yet invisible to wages. The parent raising children, the daughter caring for elderly parents, the neighbor maintaining community bonds: all contribute immense value that the labor-value nexus doesn't recognize. The automation question merely extends a problem feminist economics has already identified: the assumption that wages measure contribution was never true.

What happens to this nexus when work becomes unnecessary?

If machines can produce all we need without human labor, the economic logic for wages disappears. But the moral and social logic remains. People still need income to participate in the economy. People still need structure, purpose, identity. The system assumes work; if work vanishes, the system breaks.

Universal Basic Income: Symptom Treatment

The most commonly proposed solution is Universal Basic Income—regular cash payments to all citizens, regardless of employment status. UBI has supporters across the political spectrum: libertarians who would replace complex welfare bureaucracies with simple cash; progressives who see it as a foundation for human flourishing beyond market labor; technologists who view it as a necessary adaptation to automation.

UBI would address the immediate problem: if wages disappear, another income source is needed. It would preserve market coordination—people would still choose what to buy, price signals would still operate—while severing the link between labor and income.

We have evidence now. Finland's 2017-2018 UBI experiment found recipients happier and healthier, though not significantly more likely to find work. Kenya's GiveDirectly program has provided long-term cash transfers to villages, with evidence of improved wellbeing and local economic activity. Stockton, California's SEED program showed reduced income volatility and improved employment for recipients. The evidence suggests UBI can improve lives—but it doesn't settle whether UBI can serve as the foundation for a post-work economy.

UBI is symptom treatment, not system redesign. It answers "how do people get money?" without addressing "what is money for?" or "who decides what gets produced?" It patches the labor-value nexus rather than reconceiving it.

Consider the questions UBI doesn't answer:

Purpose: If work provides identity and structure, what replaces it? Endless leisure sounds appealing in theory but often produces anomie in practice. The deaths of despair we examined earlier struck people who had lost not just income but purpose. UBI doesn't provide purpose.

Power: If production is automated and ownership concentrated, UBI creates dependence. Citizens receive income at the pleasure of whoever controls the robots. The political economy is troubling: a small class of capital owners funding a large class of payment recipients. This is not equality; it is a new feudalism.

Inflation: If everyone receives basic income but production is constrained, prices rise and the income buys less. The real question is always about real resources, not money. UBI doesn't increase production; it redistributes claims on production. Whether that redistribution is inflationary depends on circumstances mainstream economics hotly debates.

Meaning: The deepest issue is whether market coordination remains appropriate in a post-work economy. If most production is automated, if most people don't work, why maintain the fiction of market exchange? Why not distribute goods directly, as in Fresco's vision? UBI preserves market forms without market substance—a coordination mechanism designed for a world that no longer exists.

Beyond the Wage Relation

A more radical reconception would challenge the labor-value nexus itself.

What if income was not conceived as payment for work but as a share of common wealth? The atmosphere, the electromagnetic spectrum, the accumulated knowledge of humanity—these are not products of individual labor but common inheritances. The returns on these commons could be distributed to all, as Alaska distributes oil revenues to residents, but at civilizational scale.

What if contribution was measured not in market labor but in care, creativity, community? The parent raising children, the artist creating beauty, the volunteer strengthening community—all contribute value that markets don't price. A system designed around meaningful contribution rather than wage labor might recognize and support these activities.

What if scarcity itself was challenged? If production is abundant and automation eliminates the need for human labor in production, the constraint becomes distribution, not creation. We could choose abundance—universal access to the necessities of life—rather than maintaining artificial scarcity through market mechanisms.

These are not policy proposals but thought experiments. They reveal how much our current thinking assumes the wage relation as given. Loosening that assumption opens imaginative space for alternatives.

The Coherentist Frame

From a coherentist perspective, the automation question is fundamentally about coordination.

The wage system is a coordination mechanism: it directs labor to where it's valued, distributes claims on production to those who contribute, and provides signals about what society wants produced. When labor becomes unnecessary, this mechanism breaks down. Something else must coordinate.

Markets without work become mere distribution mechanisms—ways of rationing access to what's produced without connecting that access to contribution. This may be acceptable if production is abundant and distribution is fair. But it raises the legitimacy question in acute form: why does anyone accept their allocation? By what principle is the distribution justified?

The coherentist answer is that any post-work coordination mechanism must achieve resonance—must align individual experience with collective flourishing, must provide not just goods but meaning, must earn consent rather than compelling obedience.

What that looks like in practice is not yet clear. But the question is no longer hypothetical. Automation advances. AI capabilities expand. The transformation may be gradual or sudden, but its direction seems set.

The Thread Forward

The automation question forces a reckoning with the foundations of our economic system. If labor is no longer the basis of income, if work is no longer the basis of identity, if markets are no longer the basis of coordination—what then?

The next chapter examines the coordination discontinuity more directly. What becomes possible when computation can coordinate at scales and granularities that price signals cannot? What new mechanisms are already emerging? And what are the risks when coordination power concentrates in the hands of those who control the algorithms?

For the automation question is not merely about jobs. It is about the entire system by which we organize economic life. And that system, built around assumptions that may soon be obsolete, will need to evolve—or be replaced.