The New Labour Question: AI, White-Collar Work and the Risk of a Modern Engels Moment

For much of the past two centuries, technological progress has followed a familiar pattern. Machines displaced workers, productivity increased, economies became richer, and eventually living standards improved. The transition, however, was rarely smooth. Between innovation and shared prosperity often lay decades of social upheaval.

Artificial intelligence may be pushing advanced economies into a similar period of adjustment. The difference is that the workers now most exposed are not factory labourers but members of the educated middle class.

Recent reports surrounding layoffs at major technology firms have highlighted a striking possibility: employees may be helping train the very systems that eventually reduce demand for their labour. Modern AI systems learn by observing human workflows—how programmers write code, how analysts organise information, how customer-service representatives solve problems, and how managers coordinate tasks. The objective is not merely to assist workers but increasingly to replicate portions of their work.

Whether individual companies have fully achieved this ambition remains open to debate. Yet the direction of travel is clear. The most valuable AI systems are no longer designed simply to answer questions. They are being designed to perform tasks.

A Familiar Economic Contradiction

The historical parallel is not difficult to find.

During the Industrial Revolution, Britain experienced one of the greatest productivity surges in human history. Steam power, mechanised production and expanding global trade transformed the country’s economy. National wealth increased dramatically. Yet for many workers, the early decades of industrialisation brought little immediate improvement.

The German social theorist Friedrich Engels famously documented this contradiction. Britain was becoming richer while large sections of its population remained poor, insecure and politically marginalised.

Between the late eighteenth and early nineteenth centuries, industrial output expanded rapidly. Yet real wages often lagged behind productivity gains. Urban living conditions deteriorated. Disease spread through overcrowded industrial cities. Skilled artisans displaced by mechanisation frequently failed to transition into new occupations. Many experienced long periods of hardship before broader economic gains filtered through society.

The key lesson is not that technology ultimately impoverishes societies. History suggests the opposite. Rather, the lesson is that the distribution of gains can be highly uneven, especially during periods of rapid transition.

The Industrial Revolution created extraordinary wealth. It also produced decades of disruption before institutions adapted.

AI may now be creating a similar challenge.

Automation Moves Up the Economic Ladder

Previous waves of automation primarily targeted physical and routine work.

Robots transformed manufacturing. Software streamlined clerical processes. Globalisation shifted labour-intensive production to lower-cost countries.

AI differs because it directly targets cognitive labour.

Tasks once considered uniquely human—writing reports, analysing documents, reviewing contracts, generating software code, producing marketing content and managing administrative workflows—are increasingly becoming automatable.

This does not necessarily imply mass unemployment. History suggests labour markets are more adaptive than apocalyptic predictions often assume.

The more immediate risk is subtler: fewer workers may be needed to produce the same amount of output.

A department that once required fifty employees may eventually require ten. A law firm may need fewer junior associates. A consulting company may hire fewer analysts. A bank may automate substantial portions of its back-office operations.

In economic terms, AI appears likely to increase labour productivity. The question is who captures the resulting gains.

If productivity rises faster than labour demand, the bargaining power of workers may weaken even as overall output increases.

That possibility explains why anxiety surrounding AI is concentrated not among traditional blue-collar occupations but among graduates, office workers and professionals.

The Entry-Level Problem

The greatest disruption may occur at the bottom of the white-collar career ladder.

Many entry-level jobs exist primarily because organisations require humans to perform repetitive cognitive tasks. Junior programmers write routine code. Graduate analysts prepare presentations. Junior lawyers review documents. Entry-level accountants process data.

These tasks also happen to be among the easiest for AI systems to perform.

The consequence is potentially profound.

Labour markets rely on apprenticeship structures. Workers begin with simple tasks and gradually acquire the expertise necessary for more complex responsibilities. If AI removes a significant portion of those entry-level roles, organisations may struggle to develop future generations of skilled professionals.

The danger is therefore not merely job loss but career-path erosion.

A graduate who never secures an initial foothold cannot easily accumulate the experience required for senior positions later.

Historically, technological disruption affected workers after they had entered the labour market. AI may increasingly affect workers before they enter it.

A Political Realignment in the Making

The political implications could prove significant.

The economic dislocation caused by trade shocks and industrial decline helped fuel political movements centred on globalisation, immigration and economic nationalism. Many of the communities most affected were manufacturing regions experiencing long-term job losses.

AI targets a different constituency.

The workers most exposed are often university graduates, professionals and members of the urban middle class. These groups have traditionally been among the strongest supporters of technological progress, higher education and global economic integration.

If economic insecurity spreads into these demographics, political attitudes toward technology may shift rapidly.

The traditional social contract in advanced economies has been relatively straightforward: invest in education, acquire skills and secure economic stability.

AI challenges that assumption.

For decades, education functioned as insurance against automation. Increasingly, higher educational attainment may coincide with greater exposure to it.

This reversal helps explain rising public scepticism toward AI. Many young professionals view the technology not as a source of opportunity but as a direct threat to career prospects.

The result may be a growing backlash against technology firms, corporate management and political institutions perceived as unprepared for the transition.

The Global South Faces a Different Risk

The implications extend beyond advanced economies.

Many developing countries built growth strategies around providing labour to the global economy. Outsourcing industries in countries such as India, the Philippines, Bangladesh and Vietnam created millions of jobs by supplying services and manufacturing at lower cost.

AI threatens this model.

If firms in rich countries can automate customer support, administrative processing, basic coding and routine business services, the incentive to offshore those activities diminishes.

The concern is not that outsourcing disappears entirely. Rather, the volume of labour required may decline substantially.

For countries with rapidly growing populations, this poses a serious challenge. Economic development depends on creating large numbers of productive jobs. If traditional pathways into global markets narrow before alternative industries emerge, demographic advantages may become economic liabilities.

A generation of educated young workers could find itself competing for fewer formal-sector opportunities.

This is not merely an economic problem. It is also a political and social one, potentially increasing migration pressures, public frustration and instability.

Why China May Benefit More Than Others

Not every country faces the same risks.

For ageing societies, AI may function less as a labour substitute and more as a labour replacement.

China offers the clearest example.

The country’s shrinking workforce has long been viewed as a structural constraint on future growth. AI, robotics and automation may partially offset those demographic pressures by enabling higher output with fewer workers.

In this context, automation becomes not simply a productivity tool but a demographic strategy.

China’s manufacturing base, large-scale data resources and state-directed industrial policies may allow it to capture disproportionate benefits from AI deployment.

Where younger developing countries risk insufficient job creation, ageing countries may view automation as essential to sustaining economic growth.

The same technology therefore produces very different outcomes depending on demographic circumstances.

The Distribution Problem

Technology optimists argue that these concerns are overstated.

Throughout history, innovations have destroyed old occupations while creating new ones. Agriculture gave way to industry. Industry gave way to services. Each transition generated fears that ultimately proved excessive.

There is considerable truth in this argument.

AI is likely to create entirely new industries, occupations and forms of economic activity that are difficult to predict today. Productivity gains could lower costs, accelerate growth and improve living standards.

The stronger challenge lies elsewhere.

Even if AI ultimately makes societies wealthier, it does not automatically follow that the gains will be widely shared.

This was Engels’s central observation nearly two centuries ago.

The most important question was never whether technological progress creates wealth. It usually does.

The question is who receives that wealth during the transition.

If capital owners capture most of the gains while labour absorbs most of the disruption, social tensions will rise regardless of overall economic growth.

That possibility explains why debates surrounding AI increasingly revolve around redistribution, worker protections, retraining programmes and proposals such as universal basic income.

The debate is not really about technology.

It is about political economy.

The Coming Test

The central challenge facing governments is therefore not whether AI should be developed. The economic incentives ensuring its development are already overwhelming.

The challenge is managing the transition.

History suggests that technological revolutions eventually raise living standards. It also suggests that societies can experience prolonged periods of instability before those benefits become broadly distributed.

The Industrial Revolution ultimately produced unprecedented prosperity. Yet it also generated decades of hardship, labour unrest and political conflict before institutions adapted.

AI may follow a similar trajectory.

The risk is not a world without work. The risk is a world in which productivity rises rapidly while economic institutions adjust slowly.

If that occurs, the defining political question of the coming decade may not be how intelligent machines become.

It may be how societies choose to distribute the wealth they create.