Introduction

Artificial intelligence (AI) is a strategic technology leading a new wave of technological revolution and industrial transformation, profoundly changing human production and lifestyle. General Secretary Xi Jinping pointed out that the current technological revolution and industrial transformation led by AI is in full swing. Driven by new theories and technologies such as mobile internet, big data, supercomputing, sensor networks, and brain science, AI exhibits new characteristics such as deep learning, cross-domain integration, human-machine collaboration, collective intelligence, and autonomous control, significantly impacting economic development, social progress, and global governance. While AI promotes a leap in productivity, it complicates discussions about labor forms and value creation processes, raising new questions and challenges for labor value theory that require scientific answers.

New Questions Raised by AI Development for Labor Value Theory

Labor value theory, as the foundational theory of Marxist political economy, was scientifically conceived by Marx based on the negation of classical political economy. Its main content includes theories on the dual factors of commodities, the dual nature of labor in commodity production and their interrelations, the determination and changing laws of value quantity, the development of value forms and the origin of money, and the basic contradictions and laws of commodity economy. The core idea is that value is the crystallization of undifferentiated human labor contained in commodities, and living labor is the sole source of commodity value. For a long time, there have been numerous debates in academia regarding labor value theory. In the context of rapid AI development and its profound impact on labor forms, these debates have intensified.

The Fundamental Nature of AI

Unlike traditional industrial machines, AI machines, devices, and systems exhibit characteristics similar to human intelligence, attracting widespread attention. The famous “Turing Test” aims to illustrate that if a machine can mimic human conversation to the point of being indistinguishable, its potential for intelligence should be seriously considered. The emergence of generative AI indicates that large-scale neural network models trained on deep learning algorithms can generate text, images, sounds, videos, and code, demonstrating astonishing “thinking capabilities” in the process of creating new content. This technological advancement is rapidly pushing AI toward multimodal and embodied development. Some studies suggest that AI can now achieve human-like output effects in specific tasks, primarily through learning from vast amounts of data to simulate human thought processes and generate content based on probabilistic predictions. This capability reveals intelligent characteristics distinct from traditional machines. As technology advances, AI’s behavioral performance increasingly approaches human intelligence. A profound theoretical question arises: when machine systems approach human cognitive levels, is it possible for self-awareness or independent value orientation to emerge? This question touches the philosophical boundaries of AI as a “human-like subject.” From the perspective of labor value theory, what is the elemental attribute of AI in the production system? Is it a person or a thing? Is it “human” or “human-like”?

The Subject of Intelligent Labor

In 1984, the world’s first fully automated experimental factory was built in Tsukuba Science City, Japan, bringing “unmanned factories” into public view. In recent years, with the accelerated development and widespread application of AI technology, “dark factories” have emerged rapidly. This intelligent production model relies on intelligent robots, automated devices, and digital systems to achieve full-process production, earning its name because the factory can operate in a “lights-off” state without workers present. In this context, AI exhibits a significant substitution effect for human labor. Furthermore, AI’s tireless and compliant nature can circumvent the physiological limitations, moral condemnation, and legal sanctions faced by workers engaged in long, dangerous, and intense labor, thus possessing the advantages of a “perfect worker.” Consequently, the “new labor” that Marx referred to, which emerged from the industrial revolution, will become a thing of the past; intelligent labor will become the new labor, pushing humans out of direct production processes in some industries, relegating them to roles in supervision, maintenance, and system optimization. The presence of AI in direct production and the absence of human laborers, alongside the “reappearance” of humans in research and maintenance, together constitute a complex labor landscape. This raises the question: who is the true subject of intelligent labor? Is the analysis of workers as labor subjects in labor value theory still valid?

The Source of Value Creation

According to Marx’s labor value theory, living labor is the sole source of value creation, and the value of a commodity is proportional to the amount of labor that produces it. In intelligent production systems, while AI significantly substitutes human labor, it also drives a substantial increase in social wealth and productivity levels. Some Western scholars argue that Marx’s era could not foresee a continuous production process that “does not require direct human labor intervention.” When highly automated “unmanned factories” can produce goods at nearly zero marginal cost, the source of value creation shifts from human labor to machines and algorithms themselves. Is this indeed the case? In such a scenario, has AI become a new source of value creation independent of humans? If the answer is no, then where does the value of products and services from “unmanned factories” come from? In other words, the classic proposition of labor value theory that “labor is the sole source of commodity value” requires a new response.

The Explanatory Power of Labor Value Theory for AI Applications

“From the perspective of the 500-year history of world socialism, we are still in the historical era indicated by Marxism.” As an important component of Marxist basic principles, labor value theory has not lost its theoretical explanatory power due to changes in the times; rather, its applicability has expanded, further proving its scientific nature and explanatory power. As General Secretary Xi Jinping pointed out, “Some say that Marxist political economy is outdated, that ‘Capital’ is outdated. This assertion is arbitrary.”

AI as Objectified Knowledge Power

In real life, the development and application of AI manifest in various fields and aspects. From the perspective of social reproduction, AI is also a diverse existence.

1. AI as a Product: Marx pointed out in analyzing the labor process that “labor combines with the object of labor. The object is objectified, while the object is processed. What was previously expressed in a dynamic form by the laborer now appears as a static attribute in the product.” No machine is created by nature; machines are products of human society. As an objectified product of human essential power, machines are the crystallization of past labor, knowledge, and wisdom. In this regard, AI is essentially similar to the spinning machine and the steam engine; it is created through the collaboration of scientists, engineers, and programmers, representing materialized labor with a certain value.

2. AI as Labor Material: When analyzing the changes in the capitalist labor process due to the machine system and scientific development, Marx noted that “after the production process joins capital, labor materials undergo various changes in form, with the final form being machines, or more precisely, automated machine systems… These machines are powered by automatic machines composed of many mechanical and intelligent organs.” In intelligent production systems, AI machines represent a further developed form of automatic machines, still fundamentally labor materials, albeit in a changed form. Ultimately, they are “human-created organs of the human brain,” embodying “objectified knowledge power.”

3. AI as Fixed Capital: In the social reproduction process, AI is still driven by human living labor, representing a higher efficiency of production materials. From the perspective of capital form, it also plays an important role as fixed capital in the capital accumulation process. Regarding the impact of technological progress on fixed capital, Marx pointed out long ago that “the development of fixed capital indicates how much general social knowledge has become direct productive force, and how much the conditions of social life processes are controlled and transformed by general intelligence.”

AI Cannot Replace the Subject Status of Human Labor

AI represents an enhancement of human production capacity. The manufacture and use of tools distinguish humans from animals and reflect humanity’s ability to overcome physiological limitations. From the perspective of human development history, the evolution from artificial physical capabilities to artificial intelligence represents a process of continuously breaking through human physiological limitations through technological progress, thereby achieving human liberation. The invention of machines like the steam engine allowed humans to utilize powerful machine power to replace their limited physical strength. Currently, the human-like intelligence exhibited by generative AI partially realizes the replacement of limited mental capacity. This development process reflects the continuous enhancement of human production capacity and the ongoing evolution of human liberation.

On one hand, AI lacks subjectivity in labor. From an application perspective, AI has not changed the artificial attributes and human factors in the production and research process; it cannot completely detach from human autonomous production and free growth. All of AI’s “actions” stem from preset objective functions and training data, rather than from its own survival needs or social motivations. For instance, generative AI, based on mathematical statistics and pattern matching, remains within the realm of “weak AI.” Language models can extract complex statistical patterns from vast data and achieve high-quality text generation and semantic coherence through deep learning algorithms and large-scale training, demonstrating human-like “understanding capabilities.” However, its operational mechanism fundamentally remains probabilistic modeling, incapable of performing abstract reasoning in the human sense, let alone engaging in self-awareness activities. The so-called self-iteration and self-upgrading are merely optimization processes of parameters within the framework and objectives preset by algorithm engineers, highly dependent on the quality and scale of data, with their evolutionary direction always defined and controlled by humans, which determines AI’s essential nature as a human tool.

On the other hand, this reflects the expansion of the “overall worker” in the age of AI. Marx’s analysis in “Capital” indicates that the concept of the “overall worker” expands from handicraft industry to large-scale industrial machinery. The various members of the overall worker directly or indirectly act on the labor object. Therefore, as the collaborative nature of the labor process develops, the concepts of productive labor and its bearers—the productive workers—must also expand. In the age of AI, as social production’s division of labor and collaboration further develops, the concept of the overall worker inevitably expands beyond the traditional factory walls. Essentially, “unmanned factories” are not truly without workers; they are merely a part of the entire production process, still backed by numerous researchers, annotators, and operational personnel engaged in algorithm training, data annotation, hardware maintenance, and other tasks. The laborers who once monitored and debugged machines have now transformed into those organizing data, analyzing data, and issuing data commands to machines; they are the true subjects of intelligent labor.

AI is Not a Source of Value

AI machines not only directly participate in commodity production but also displace workers, creating the illusion of AI dominating commodity production and creating value. In reality, AI machines are products of materialized labor; they “produce nothing as value” and cannot become sources of value. Throughout the production process, AI lacks complete autonomy and cannot completely detach from humans, nor can it create value without human living labor. Regarding generative AI, regardless of how novel or creative the content it “produces” may seem, it is merely a probabilistic deduction and output based on existing program designs and algorithm models, and this computational process lacks social attributes. It must be activated and realize value transfer through human input, goal setting, and process intervention. The value creation in AI production systems stems from the living labor of humans engaged in intelligent production, originating from the intelligent labor of the overall worker. Why does human living labor seem to decrease while the value of products produced by some AI companies does not diminish but rather increases? Generally, there are three main reasons:

1. Value Created by Intelligent Workers’ Living Labor: “Labor exists not as an object but as an activity; it exists not as value itself but as the living source of value.” In intelligent production systems, value originates from the living labor of those operating, maintaining, and managing AI machines during production. For generative AI, living labor primarily includes the complex intellectual labor of computer scientists, data engineers, and algorithm engineers continuously optimizing models; it also encompasses the relatively mechanical and fragmented tasks of “digital laborers” engaged in data collection, annotation, image segmentation, semantic cleaning, and content review, which condense vast amounts of undifferentiated human labor into underlying data. Together, these forms of living labor constitute the true source of value for AI products.

2. Value Transfer of Constant Capital: The value transferred to new products includes portions from production raw materials and auxiliary materials, as well as the depreciation of AI machines. A large number of researchers invest substantial labor in the development of new model architectures, algorithm iterations, and computational system deployments and maintenance, which condenses value into AI machines. As labor materials, AI machines will “transfer their value to the products produced by their services, just like any other component of constant capital.” The value added in production will never exceed the value lost due to wear and tear. This depreciation includes both tangible wear caused by usage and idleness and intangible wear due to technological iteration.

3. Redistribution of Profits: Marx’s analysis in “Capital” indicates that enterprises that adopt advanced technologies and improve labor productivity will obtain excess profits. It is evident that, for a certain period, a few AI technology companies leverage their advantages in data, algorithms, and computational power, resulting in their individual labor productivity being significantly higher than the industry average, thereby forming technological barriers, gaining market dominance, and obtaining excess profits. This essentially represents a redistribution of profits among capital.

Growth Points of Labor Value Theory in the Age of AI

Compared to the time when Marx established labor value theory, significant changes have occurred in today’s era. “Marxist political economy must be dynamic and keep pace with the times.” We must seriously focus on and study the theoretical growth points regarding changes in labor forms and value creation spaces, enriching and developing Marxist labor value theory in a timely manner.

Major Changes in Labor Forms

As a unique conscious and purposeful social practice activity, labor is the material exchange process in which humans consume their physical and mental energy to regulate the relationship between humans and nature. Currently, the rapid development and widespread application of AI have given rise to intelligent labor, a new form of human labor. Compared to the presence, repetitiveness, and specificity of labor forms in the mechanical industrial era, intelligent labor exhibits prominent characteristics of absence, creativity, and abstraction, achieving an iterative upgrade of human labor forms. Marx noted that “to engage in productive labor, it is no longer necessary to do it personally; it is sufficient to become an organ of the overall worker and fulfill a certain function.” With the development of intelligent labor, the concepts of production workers and productive labor have further expanded.

In intelligent production systems, humans and AI form a complementary and collaborative relationship, with AI undertaking repetitive, high-intensity, high-precision, and high-risk programmed tasks, while humans focus on creative labor. Data has become an important production factor, with many people engaging in “digital labor,” which exhibits characteristics of non-material forms. The production and circulation of data have spawned a series of new industrial branches, including a complete industry chain dedicated to data collection, annotation, cleaning, classification, storage, analysis, and contextual application. Data can be infinitely replicated and artificially deleted, unlike traditional use values that possess specific and limited utility; thus, “digital labor” is difficult to quantify in terms of time. It is essential to study the new definitions of productive labor brought about by “digital labor”; to examine the new characteristics of intelligent labor; to reassess the distinctions between mental and physical labor, complex and simple labor, and the transformation of private labor into social labor, and to provide new interpretations of labor quantity calculations. For the political economy of socialism with Chinese characteristics, it is crucial to study how to promote the integration of the digital economy and the real economy, leveraging digital technology to amplify, overlay, and multiply economic development.

Expansion of Value Creation Space

In intelligent production systems, production activities exhibit significant characteristics of fragmented labor time, virtualized labor management, and flexible labor locations. Intelligent labor extends from factory workshops to digital platforms and social networks, greatly expanding the space for value creation. Some foreign scholars have proposed the concept of “platform capitalism,” indicating that public network platforms serve as new platforms for value creation, with value primarily derived from the extraction and capitalization of user behavior data, rather than being limited to traditional compensated wage labor. Users’ attention, clicks, browsing, and comments on digital platforms are recorded and used by the platform to optimize algorithms and customize more precise products and services. In this regard, the global “digital laborers,” including workers producing hardware, software development engineers, data processing personnel, and users generating data through online clicks, collectively create value in various forms. Among them, the “digital labor” provided by users on major online platforms is often unpaid, passive, and not covered by traditional value theories, complicating the identification of value sources. Therefore, it is necessary to expand the categories of “labor,” “value creation,” and “laborers” to encompass the new phenomena of the AI era and deepen the understanding of the new characteristics of value creation.

New Characteristics of Labor Process Control

Labor value theory not only profoundly reveals the basic principles of commodity value formation but also lays the theoretical foundation for uncovering the sources of surplus value. Under the widespread application of AI, control over the labor process no longer primarily relies on management supervision by foremen or tangible production discipline constraints but is instead achieved mainly through systems governed by data and algorithms, exhibiting greater concealment, internalization, and real-time characteristics. Research on gig economy platforms in academia has shown that AI algorithms have achieved unprecedented fine control over the labor process through order dispatching, pricing, supervision, and rating, making labor process control more “exquisite” and creating a so-called landscape of production without participation or exploitation. In reality, “digital labor,” unrestricted by time and space, can harm workers’ physical and mental health, while continuously optimized algorithms can alienate labor into negative activities of “physical torment and mental destruction.” Engels once said that as human labor capacity continuously grows, “science increasingly makes natural forces subject to human control. This immeasurable productive capacity, once consciously applied for the benefit of the masses, will quickly reduce human labor to a minimum.” To realize Engels’ vision, the key lies in shifting the purpose of technological development from “value proliferation” to “human liberation.” In this regard, the contemporary mission of labor value theory is to reveal the roots of technological alienation under capital logic and provide theoretical support for technology to benefit the masses.