Does AI Change the Labor–Capital Relationship, or Merely Make It More Efficient?

A new machine does not automatically create a new social relationship

Artificial intelligence appears to change the relationship between labor and capital because it can perform tasks once associated with human judgment, language, research, design, and coordination.

It can write, classify, predict, recommend, summarize, and generate. It can operate continuously and reproduce a form of performance across thousands of users.

But technical novelty and social novelty are not the same thing.

The important question is whether AI changes who owns productive capacity, who controls the organization of work, who receives the gains, and who carries the risks. If those relationships remain unchanged, AI may be a new technical instrument serving an older economic structure.

Marx: machinery inside the labor–capital relationship

Karl Marx’s analysis of machinery provides a useful starting point because it does not treat technology as an independent historical force. A machine can increase the productive power of society, but its social consequences are shaped by the ownership system in which it operates.

Under private control, greater productive capacity can become a means of reducing labor costs, intensifying work, and increasing the surplus available to owners. The machine changes how production happens without necessarily changing who commands production.

AI may follow the same pattern. It can increase the output of a worker while the organization retains control over the tool, the data, the workflow, and the productivity dividend.

In that case, AI changes the instrument of production more clearly than it changes the relationship between those who work and those who own.

Thompson: AI can reorganize time at a deeper level

E. P. Thompson’s account of industrial time helps reveal what AI adds to this older pattern. Industrial systems turned time into a measurable and enforceable structure. Work became synchronized with clocks, schedules, machines, and production targets.

AI can extend this control into knowledge work. A system can monitor queues, recommend priorities, measure response times, assign tasks, and generate new expectations in real time.

The worker is not only asked to complete a task. The worker is placed inside a continuously adjusting temporal system.

AI may therefore make the labor–capital relationship more efficient by reducing the distance between performance measurement and managerial intervention. The organization can observe, compare, and revise work at a speed that older management systems could not match.

Braverman: when management enters the workflow

Harry Braverman’s analysis of work focused on the movement of knowledge from workers into management systems. The more an organization can separate planning from execution, the more it can standardize and control the work process.

AI may intensify this movement. It can encode preferred methods, generate instructions, recommend decisions, and evaluate outputs. Part of the worker’s practical judgment becomes a feature of the system.

This can support workers when the system expands their abilities and leaves them with meaningful discretion. It can weaken them when the system defines the correct process, measures compliance, and treats professional judgment as an inefficient deviation.

The worker may still be present, but the authority to decide how work should be done has moved into the software.

Where AI genuinely changes the relationship

AI is not only another machine. It introduces changes that can alter the structure of labor in important ways.

These differences matter. AI may not simply replace a task. It can change how expertise is formed, how workers enter professions, and how responsibility is assigned.

If beginner tasks are automated before beginners have time to learn them, the training ladder can become compressed. Organizations may receive immediate efficiency while weakening the future supply of experienced workers.

Gray, Suri, and Casilli: the labor AI continues to need

Mary L. Gray and Siddharth Suri’s work on ghost work shows how automated services can depend on people performing fragmented tasks behind the interface. Antonio Casilli’s research on digital labor similarly emphasizes that artificial intelligence reorganizes human work rather than simply eliminating it.

Data preparation, labeling, evaluation, moderation, correction, and exception handling remain part of the system’s operation. The output appears automated because this labor has been distributed, concealed, or placed outside the user’s view.

AI therefore changes the location and visibility of labor. It does not prove that labor has disappeared.

Crawford: AI is also a material system

Kate Crawford’s analysis of artificial intelligence adds another necessary dimension. AI is not only a set of algorithms. It depends on energy, minerals, data centers, supply chains, engineering, maintenance, and workers distributed across different parts of the world.

The apparent intelligence of a system can therefore hide both human labor and material extraction.

A model may answer a question in seconds, but that speed does not describe the full production system required to make the answer possible. If the account measures only the final response, the infrastructure and labor supporting it appear to be free or naturally available.

Pasquinelli: whose intelligence is being organized?

Matteo Pasquinelli’s work encourages us to ask whether machine intelligence is independent intelligence or a technical reorganization of collective intelligence.

Language, classification, design, problem-solving, and professional judgment are not created by isolated individuals. They accumulate through social practice. AI systems can gather patterns from that accumulated activity and reorganize them into a productive instrument.

The system may then appear to own the intelligence it has absorbed.

This does not make AI identical to human labor. It does make the question of ownership more difficult. If collective intelligence becomes a privately controlled productive asset, the labor–capital relationship may expand from the workplace into the knowledge commons itself.

More efficient continuity or genuine transformation?

AI continues older patterns when ownership remains concentrated, workers have little say over deployment, productivity gains become new expectations, and human labor is hidden behind automated outputs.

It creates a deeper transformation when the object of control shifts from physical activity to knowledge, judgment, data, and collective intelligence. The system can influence not only what workers do, but also how they learn, how they become experts, and whether their contribution remains recognizable.

The two descriptions can be true at the same time. AI can be historically new and economically familiar.

The test is not intelligence but power

Calling a system intelligent does not tell us who controls it. A model may be technically impressive while operating inside a conventional hierarchy of ownership and command.

The more decisive questions are practical:

• Who owns the model, data, and infrastructure?
• Who decides how AI changes the workday?
• Who receives the productivity gains?
• Who becomes responsible when the system fails?
• Who has the right to refuse, modify, or govern its use?

If the answers remain concentrated on the capital side of the relationship, AI may be transforming production while making the existing structure more efficient.

TravelIAQ Smart Tip: To determine whether AI has changed a labor–capital relationship, look beyond what the system can do. Examine ownership, control, distribution, training, accountability, and worker voice. A new capability becomes a new social relationship only when power over that capability is redistributed.