Blog Post: Technological Anxiety Is Not New: What Past Labor Disruptions Teach Us About AI

This post is the second in a series titled “AI and the Future of Work,” hosted on the Notre Dame-IBM Technology Ethics Lab’s blog. In this series, we will examine the implications of AI for jobs and society: how tasks and roles might change, how organizations can adapt, how workers are affected, and what governance frameworks are being considered to create and support beneficial and responsible outcomes.

Why this series turns to history

This blog is the second installment in “AI and the Future of Work,” a series that examines how contemporary AI systems, especially generative and increasingly agentic tools, may reshape labor, employment, and workplace governance. The goal of this series is not to treat “the future of work” as fixed or inevitable, but to clarify what is changing, what is not, and what can be shaped through institutions, organizational choices, and public policy research.

A useful starting point is a deceptively simple observation: large technological transitions almost always produce labor anxiety. The reasons are understandable. New technologies frequently (1) substitute for human effort in specific tasks, (2) reorganize how work is coordinated, and (3) shift bargaining power and wages, often before societies have adapted rules, training systems, or social protections to match the new reality.

These dynamics were visible in the early industrial era (textile machinery and factories), the mechanization and mass-production era (tractors, electrification, and assembly lines), the computer era (routine-task automation and occupational polarization), and the internet era (reconfigured markets and “winner-take-more” dynamics tied to connectivity and intangible assets).

The historical record, as this Blog will discuss, does not suggest that “technology always destroys jobs” or that “technology always creates better jobs.” Instead, it suggests:

• The most consequential effects tend to be distributional (who gains, who loses, how quickly, and under what protections).

• Productivity gains often require complementary investments (skills, organizational redesign, infrastructure, and regulation), and these complements can take years or decades to develop.

With that framing, the remainder of this post reviews four major technological transitions—industrialization, mechanization, computers, and the internet—and considers how scholars draw on these lessons to inform today’s “future of work” discussions, including a central question raised in ethics literature: what would it mean for AI adoption to be efficient while also supporting fairness and human well‑being?

Industrialization: the factory system and the fear of machines

One of the clearest early examples of technological anxiety comes from the early nineteenth century, where skilled textile workers saw mechanized production as a direct threat to wages and livelihoods. The Luddite movement is often caricatured as “anti-technology,” but many historical treatments emphasize that Luddite actions were closely tied to economic insecurity, labor conditions, and the perceived use of machines to undercut wages and customary work standards. The British state’s response illustrates both the intensity of the conflict and how quickly labor disputes can become governance disputes. The 1812 “Frame Breaking Act” made certain forms of machine-breaking a capital offense and the UK’s National Archives report that 17 executions are connected to the unrest in the period. What happened to labor outcomes after early industrialization? Industrialization ultimately increased productivity and transformed economic structures, but the benefits were not immediate or evenly shared. Economic historians have documented a period often summarized as “Engels’ Pause,” in which output per worker rose substantially while real wages rose much more slowly.

Mechanization: electrification, assembly lines, and the long shift out of farm work

If early industrialization made “machinery” a symbol of labor threat, the later mechanization era made something else visible: the fact that technology changes work not only by replacing tasks, but by reorganizing production systems. This point is central to classic scholarship on electrification and productivity. In his influential historical analysis, Paul David argued that major general-purpose technologies (using the dynamo and electricity as a reference point) often require extensive complementary changes such as factory layouts, workflows, managerial practices, and skills, before productivity gains appear at scale.

Mechanization’s workplace face is often associated with mass production and the assembly line. But mechanization also intensified concerns about job quality: repetitive tasks, strict pacing, and the “discipline of the line.”

A striking data point from archival description is that labor turnover at Ford reportedly reached extremely high levels in the early assembly-line period, which reflects the human costs of work reorganization. The company’s “Five Dollar Day” policy (1914) is often interpreted as one response to retention and labor stability pressures in high-discipline production settings. Whether one views this as generosity, managerial strategy, or both, it shows that work reorganization triggers institutional responses like bargaining, not just technical efficiency.

Mechanization’s policy responses in the U.S. highlight a second major “institutional lesson”: social insurance and labor standards expanded during periods of industrial and economic transformation. For example, the Social Security Act of 1935 enabled state unemployment compensation systems among other provisions, and the Fair Labor Standards Act re-shaped work hours and compensation norms.

Mechanization reinforced that technological transitions are not “just about machines.” They are about systems—infrastructure, work design, labor standards, and social insurance. Whether technology produces widespread benefits depends heavily on all of these complements.

Computers: “technological unemployment,” routine work, and job polarization

Anxiety about machines displacing labor did not end in the nineteenth century. It reappeared explicitly in twentieth-century economic thought. John Maynard Keynes famously used the term “technological unemployment” in a 1930 essay, describing a scenario where labor-saving innovation could temporarily outpace the creation of new uses for labor. While Keynes was writing in the context of interwar economic uncertainty, the phrase became newly salient during waves of automation talk later in the twentieth century and remains a touchstone in current AI discourse.

The computer era’s most policy-relevant labor shift is not simply that “computers automated work,” but which work they automated. Task-based research argues that computer capital tends to substitute for routine cognitive and manual tasks that can be codified as explicit rules, while complementing non-routine problem-solving and complex communication tasks.

One widely discussed consequence of the rise of computers in the workplace is labor market “polarization”: growth in high-wage, high-skill jobs and low-wage service jobs, paired with relative decline in many middle-skill routine occupations. Econometric studies using U.S. labor market data document an increase in wage inequality trends and connect employment polarization patterns to computerization’s differential effects across task categories.

At the individual level, early empirical work also found that workers who used computers at work earned a wage premium, and the diffusion of computer use was analyzed as a potential contributor to rising returns to education in the 1980s. But the computer era also produced methodological caution: researchers pointed out that computer use can be correlated with workplaces and jobs that are already higher-paying for other reasons.

Finally, computerization showed again that technology’s benefits are not automatic; they depend on organizational complements. Research shows that the value of IT investments is closely tied to complementary organizational transformation, change management, and intangible investments.

The internet: connectivity, uneven payoffs, and new governance agendas

The rise of the internet produced a different but related form of technological anxiety: concerns about disintermediation, online retail disruption, new forms of competition, and geographic inequality as markets became more connected and information costs fell. The labor effects of the internet era are often best understood as accelerating the importance of digital infrastructure and amplifying the role of intangible, scalable assets such as software, data, and networks.

A key empirical finding from research on regional effects is that while internet diffusion and exposure was broad, wage gains were not uniformly distributed. In one influential study of U.S. counties, advanced internet technology investment was associated with substantial wage growth only in a small subset (around 6%) of already-advantaged counties.

The internet era also made market structure a central governance concern, including antitrust attention to platform and software ecosystems.

What these histories teach us about AI and associated technological anxiety today

Current AI anxiety often sounds familiar because it recombines earlier patterns: fear of displacement, uncertainty about skills, and distrust that gains will be fairly shared. Contemporary survey and policy documents capture the intensity of these concerns; for example, OECD reporting on workplace AI notes substantial worker worry about job loss and wage effects, even while many workers and employers also report positive impacts on tasks and performance.

At the same time, AI is not simply “the next computer” or “the next internet.” The historical comparisons are most useful when they clarify both continuities and differences.

Continuities that history makes hard to ignore

• Task disruption is more predictable than job extinction. Just as textile machinery, tractors, and computers substituted for some tasks while complementing others, much AI analysis now distinguishes between occupational “exposure” and real-world outcomes such as automation, augmentation, and job redesign. The International Labour Organization’s analysis of generative AI emphasizes that exposure does not mechanically translate into full automation and anticipates that augmentation effects may dominate in many settings.

• Distribution matters as much as aggregates. The industrial era’s “Engels’ Pause” and the computer era’s polarization show that labor anxiety can be rooted in the lived experience of stagnating wages, degraded job quality, or eroded bargaining power, even when output and productivity rises. Current policy analysis similarly warns that AI can deepen inequality if benefits accrue primarily to owners of capital, highly skilled workers, or already-advantaged regions and firms.

• Complementary investments take time. Electrification and IT both illustrate that productivity gains depend on organizational redesign and intangible complements such as training, new workflows, and process reengineering. This is directly relevant to AI: if organizations deploy generative or agentic systems without redesigning work, setting quality controls, or training staff, they should not expect stable productivity gains, or fair outcomes, by default.

Differences that make AI governance distinct

• AI supports cognitive and communicative tasks at scale. Computers strongly affected routine tasks, but generative AI systems are designed to perform or assist with a broader range of language-mediated activities (drafting, summarizing, translating, ideation) and, in more agentic configurations, can plan and execute multi-step workflows. This expands the set of occupations potentially affected, including many white-collar roles that historically felt more insulated.

• Speed and diffusion dynamics may be faster. As with the internet, software-based technologies can spread quickly once integrated into platforms and workflows. AI diffusion may therefore compress the adjustment timeline for workers and institutions.

A Historical Snapshot

The table below is designed as a quick reference: it compresses the “anxiety → outcomes → responses → lessons” pattern across eras.

Perla Khattar is a J.S.D. candidate at the University of Notre Dame Law School and a Research Fellow at the Notre Dame-IBM Tech Ethics Lab. In her research, she analyzes emerging technologies such as synthetic data, neurodata, quantum computing, and AI through a privacy lens, focusing on how law and policy can anticipate technological disruption while safeguarding individual rights.