When a single research note from an independent firm can help erase $200 billion in software stock value, fear has become a tangible market force. The announcement by Block CEO Jack Dorsey that 4,000 workers—nearly half its staff—would be cut because artificial intelligence has “changed what it means to build and run a company” gave this anxiety a name. It is the Fear Of Becoming Obsolete (FOBO), and it is moving from theoretical discussion to real-world consequence.
The market’s reaction is a barometer of this sentiment. Yet this panic, while understandable, is misdirected. The critical question is not whether AI will destroy jobs in the aggregate. It is about who captures the gains from the productivity it unlocks, and who bears the costs of the transition. The historical record and an honest assessment of capital incentives suggest the primary outcome of this technological wave will not be mass unemployment, but a deepening of economic inequality under existing power structures.
Technological change has never produced a permanent job apocalypse. To believe this time is different is to ignore the fundamental dynamics of a capitalist economy. It is an economy that grows. It is not a zero-sum game with a fixed number of jobs to be divided. This analysis requires a dispassionate look at the mechanics of automation, both past and present.
The Engine of Disruption is Profit Not Technology
Automation is the replacement of human tasks with machines. This process has two primary forms. The first replaces an entire worker, rendering a profession obsolete, as automated spinning did to nineteenth-century textile spinners. The second replaces a specific tool or task, augmenting the worker, as the electric drill augmented the technician. Both increase productivity, which is the ability to produce more output with the same, or less, labor input. This is the rational basis for the fear of job loss.
Yet, the capitalist is not primarily motivated by a desire to control labor or even, necessarily, to eliminate it. The capitalist is a prisoner of the market. The overriding imperative is profit, driven by relentless competition that forces firms to lower the unit cost of their goods. Technology is the primary weapon in this war. A firm introduces new machinery or software to increase efficiency and undercut competitors. Control over the labor process is a means to this end, not the end itself. If making work easier or more pleasant maximized profit, it would become the norm. (It is not.)
This dynamic creates two simultaneous effects, as Karl Marx observed centuries ago. Technical change displaces labor, pushing workers into a “reserve army of the unemployed.” But that same change also accelerates economic growth. Increased productivity means the economic pie expands. As firms become more efficient, they can lower prices or invest their higher profits, stimulating demand. New plants open. New services are created. This expansion creates a demand for labor, absorbing those who were displaced.
This is not theory. It is observable history. The introduction of the Automatic Teller Machine (ATM) was predicted to wipe out the bank teller profession. Instead, from the 1990s to the early 2000s, as ATMs proliferated, the number of bank tellers actually rose. ATMs increased bank profitability, allowing them to open more branches. The role of the teller simply changed, shifting from clerical cash-handling to customer relations and financial services sales. The task was automated; the job was redefined.
General Purpose Technology and Historical Precedent
The argument that AI is qualitatively different rests on its status as a general-purpose technology (GPT), an innovation that fans out across the entire economy, unlike a specialized tool. Electricity and the computer are prior examples. AI certainly fits this description. Its potential applications are vast, not confined to a single sector. This may mean its effects are more widespread, but it does not mean they are without precedent.
We have seen GPTs before. Each wave—steam power, electrification, computerization—was accompanied by the same apocalyptic predictions. Each time, the economy adjusted. Jobs were lost in specific sectors, but new occupations emerged and overall unemployment rates remained broadly stable, oscillating with the business cycle, not on a permanent upward trajectory.
Furthermore, the current capabilities of AI are largely an extension of what computers already do: executing clearly defined, replicable tasks based on explicit knowledge. It excels at building spreadsheets, performing accounting functions, and writing code. So far, its inroads into work requiring “tacit knowledge”—the judgment, instinct, and creativity that humans acquire through experience—have been minimal. Instead of replacing coders and accountants, it is primarily being used as a tool to make them more productive. It is augmenting, not eliminating. This may change, but prognostications of AI creating new AI in a recursive loop of job destruction remain in the realm of science fiction.
The Real Crisis Distributional Consequences
The macroeconomic view that aggregate employment will remain stable is cold comfort to the individual worker displaced by automation. The costs of this transition—loss of income, community, and identity; the financial and emotional burden of retraining and relocating—are borne almost entirely by the individual. The benefits, however, flow elsewhere.
This is the central issue the current panic obscures. The problem is not technology; it is the distribution of its rewards. Since the mid-1970s, a crucial link in the economic engine has been broken. For a century prior, from roughly 1860 to 1970, wages and productivity in the United States moved in tandem. As workers produced more, they earned more. Around 1975, these lines diverged sharply. Productivity continued its upward climb, while real wages for the majority of workers stagnated.
This great decoupling coincides with two structural shifts. The first was the precipitous decline of the trade union movement, which eroded the institutional power of labor to bargain for a share of productivity gains. The second was the acceleration of globalization, which allowed capital to play domestic workers against a global labor pool, exerting relentless downward pressure on wages.
AI enters an economy already defined by this imbalance of power. Without countervailing forces, it will function as an accelerant. The increased profits generated by AI-driven efficiency will accrue to capital owners and a small class of highly skilled professionals. The broader workforce, lacking bargaining power, will see little of this gain. The result is a politically mediated outcome, not a technologically determined one.
A Future of Intensification and Inequality
The immediate future of work shaped by AI is not one of leisure, but one of intensification. For those whose jobs are not eliminated, AI tools will be used to expand their scope of work, leading to increased multitasking, cognitive fatigue, and workload creep. An accountant will not work less; they will be expected to handle more clients. A programmer will not go home early; they will be expected to ship more code. (The notion that CEOs see saved hours translating directly into reduced headcount has already been questioned by executives like JPMorgan’s Jamie Dimon.)
Simultaneously, a structural crisis is brewing at the entry-level. As a Dallas Fed report noted, AI is adept at handling “textbook knowledge” tasks, which have historically been the training ground for new workers to develop the “street smarts” of experience and tacit knowledge. If AI handles the grunt work, the ladder for developing the next generation of experienced professionals is kicked away. This creates a long-term sustainability problem for businesses and a bottleneck for young workers already facing a deteriorating job market.
To navigate this, the focus must shift from blocking technology to building social shock absorbers. A robust welfare state, generous unemployment insurance, and ambitious public retraining programs are not socialist fantasies; they are pragmatic necessities to manage the social costs of a transition that benefits the economy in aggregate. They socialize the costs currently borne by individuals.
The debate should not be about a hypothetical job apocalypse. It should be about political and economic power. The trajectory of AI’s impact will be determined by the strength of organized labor and the willingness of the state to intervene. Left to its own devices, capital will deploy AI to maximize profit, intensify work, and deepen inequality. Technology has the potential to reduce human toil, but within the current system, its primary function is to serve the bottom line. That is the real, and far more urgent, crisis.