Can Human Labor and AI Labor Be Exploited Through the Same Logic of Maximum Output and Minimum Cost?

The same logic does not create the same experience

Human labor and artificial intelligence are not the same kind of thing. A person has a body, needs rest, receives or seeks a wage, develops a life outside production, and can experience pressure, insecurity, exhaustion, and injustice.

An AI system has none of these human conditions. It does not have a household, a biological limit, or a subjective experience of being overworked.

Yet human labor and AI capability can still be placed inside the same economic logic.

An organization can ask of both: How much output can be produced? How quickly can it be produced? How little can it cost? How continuously can the system operate? How much of the resulting value can the owner retain?

The shared logic is real, even though the two forms of production remain fundamentally different.

Marx and the logic of extraction

Karl Marx’s analysis of labor and machinery helps clarify the comparison. Human labor becomes part of a capitalist production system when labor power is purchased, directed, and used to create more value than the worker receives in return.

Machinery enters the same system as a means of increasing productivity, reducing costs, and expanding the surplus available to the owner. The machine is not a worker in the human sense, but it can become an instrument for organizing and extracting human labor.

AI extends this arrangement into activities traditionally associated with knowledge and judgment. It may increase the output of one worker, reduce the number of workers required, or allow a company to serve more customers without proportionally increasing wages.

The central economic relationship remains recognizable: productive capacity is privately controlled, while the people who operate, support, or contribute to that capacity do not necessarily control its distribution.

The common operating logic

Thompson and the ownership of time

E. P. Thompson’s account of industrial time shows how production systems turn time into a resource that can be measured and controlled.

For human workers, maximum output can mean tighter schedules, fewer pauses, and the conversion of saved time into additional assignments. AI changes the scale of this process by making certain forms of assistance available continuously.

Once an AI system can draft, search, summarize, or respond at any hour, organizations may begin to treat that availability as the new normal. The worker is then expected to answer faster, handle more cases, and remain connected for longer.

AI does not become tired. Human workers still do. A system that never rests can nevertheless create a workplace that expects people to behave as if they never need rest.

Braverman and the control of knowledge

Harry Braverman’s analysis of management focused on how practical knowledge can be removed from workers and reorganized inside systems controlled by the organization.

This helps explain why AI can change the status of knowledge work. A worker’s methods, examples, corrections, and judgments may be collected and embedded in software. The organization then gains a reusable capability that can be applied beyond the original worker.

The worker may become more productive with the tool, but the organization may also become less dependent on the worker’s individual knowledge.

This creates a double movement. AI can extend a worker’s ability while weakening the worker’s bargaining power if the system captures the knowledge and the owner controls the resulting capability.

Gray, Suri, and Casilli: the human labor inside AI labor

Mary L. Gray and Siddharth Suri’s work on ghost work shows why the phrase “AI labor” can be misleading when it hides the people supporting the system.

Human workers may prepare data, label examples, evaluate responses, moderate content, correct mistakes, and handle cases that the system cannot resolve. Their work is often divided into small tasks and placed outside the visible product.

Antonio Casilli’s research on digital labor similarly demonstrates that automation frequently reorganizes human work rather than eliminating it. The system appears to produce an answer independently, while human labor remains distributed throughout the pipeline.

If those workers are poorly paid, invisible, or excluded from decisions about the system, the organization may present AI as cheap automation while shifting the real burden onto a less visible workforce.

In this sense, the exploitation is not experienced by the AI. It is experienced by the people whose labor makes the AI appear autonomous and inexpensive.

Crawford and the material limits of “free” intelligence

Kate Crawford’s analysis of AI adds the physical world back into the discussion. AI systems require energy, minerals, data centers, cooling, manufacturing, engineering, maintenance, and global supply chains.

These resources are not free simply because they are absent from the user interface.

A system may process millions of requests while the costs of infrastructure and environmental impact remain distributed across communities and future users. Maximum output can therefore be pursued by hiding the material conditions that make the output possible.

The same logic that minimizes the visible cost of labor can minimize the visible cost of energy, extraction, and infrastructure.

Pasquinelli and collective intelligence

Matteo Pasquinelli’s work helps explain why the comparison extends beyond individual workers. AI systems are built from patterns of language, classification, design, problem-solving, and judgment accumulated through collective activity.

When these capabilities are reorganized inside a privately controlled model, collective intelligence becomes a productive asset that may be used without collective ownership.

The system appears to provide intelligence at low cost because the social history behind that capability is no longer visible at the moment of use.

This does not mean that AI is literally a collective human worker. It means that the productive capability of AI has social origins, and those origins matter when asking who should benefit from its expansion.

Where the analogy breaks down

The phrase “AI labor” becomes dangerous if it erases the difference between a person and a technical system.

A human worker can be exploited through wages, working time, insecurity, bodily exhaustion, and the loss of control over life activity. An AI system cannot experience these conditions in the same way.

When people say that AI is being exploited, they may be describing the use of a system beyond its intended limits, the extraction of value from its outputs, or the underpayment of the human labor and infrastructure that support it.

The more precise question is therefore not whether AI suffers exploitation. It is whether AI is being used as part of an economic arrangement that exploits people, appropriates collective knowledge, and externalizes material costs.

What changes when both are optimized together?

When human workers and AI systems are placed inside the same production process, the organization may compare them through a common metric: output per unit of cost.

This can create pressure on both sides of the system. Human workers are compared with automated capabilities, while AI systems are supported by human workers whose labor is expected to remain flexible, cheap, and invisible.

The organization may use AI to reduce human labor costs while depending on human labor to train, supervise, correct, and legitimize the AI.

This produces a circular arrangement: people make the system more capable, the system makes some people more replaceable, and the resulting savings are used to demand further efficiency.

The real question is who controls the combined system

Human labor and AI capability can be governed by the same logic of maximum output and minimum cost. That does not make them identical. It reveals the power of the institution that organizes them together.

The decisive questions are:

• Who owns the AI system and its infrastructure?
• Who contributed the knowledge and data from which it learned?
• Who receives the productivity gains?
• Who carries the risk when the system fails?
• Who decides whether efficiency becomes freedom or another demand?

If the answers remain concentrated among owners and managers, AI may become a new instrument for applying an old production logic to a wider range of human activity.

TravelIAQ Smart Tip: When comparing human work with AI output, do not compare only speed and cost. Include training, maintenance, supervision, correction, infrastructure, energy, responsibility, and social consequences. Maximum output at minimum visible cost may simply mean that the hidden costs have been transferred elsewhere.