Zhu Hongwei: "Human-AI Collaboration" and the Transformation of Labor Forms

The State Council's Opinions on Deeply Implementing the "AI Plus" Initiative propose to "accelerate the formation of new patterns of an intelligent economy and intelligent society characterized by human-machine collaboration, cross-sector integration, and co-creation and sharing." The deep application of Large Language Model (LLM) technologies is shifting artificial intelligence from a tool of labor into a "collaborative subject," reshaping the existing forms and relational modes of labor. This transformation not only promotes "human-machine reconciliation" through technical progress—liberating laborers from simple, repetitive labor to focus on creative, high-order labor—but also facilitates the penetration of the logic of capital due to high capital and knowledge requirements, bringing about a series of intertwined systemic risks. Therefore, it is necessary to return to the "hidden abode of production" ^[1] to uncover the obscured labor processes and production scenes of the era of "human-intelligence collaboration." On the basis of preventing and resolving risks, we must promote the construction of labor relations characterized by human-intelligence symbiosis and co-evolution.

Collaboration of Labor Power and New Forms of Human-Machine Cooperation. The application of intelligent large models in production continuously drives the "intelligent emergence" and performance optimization of machines through the exponential growth of computing power and the explosion of data. Intelligent machines have acquired increasingly powerful and diverse "labor power" ^[2], exhibiting brand-new characteristics of intersubjectivity. The human-machine relationship in traditional production is shifting from a master-slave division of labor mode of "human instruction—machine execution" to an equal collaboration mode of "human leadership—intelligent assistance." Intelligent machines are no longer purely passive tools of execution but have become "partners" actively collaborating in human labor. This change is not a simple iteration of technical tools or cooperative modes, but a systematic restructuring of the subjects of labor, value creation, and relations of production. As Marx pointed out, "the collective character of the labour-process presents itself as a technical necessity dictated by the instrument of labour itself."

First, the labor form of human-intelligence collaboration continuously integrates the respective labor advantages of humans and machines, opening up a "missing middle ground" that requires human-machine cooperation. Generally speaking, intelligent machines excel at processing and analyzing data and performing repetitive tasks, while complex human cognitive abilities are suited for dealing with complex and unpredictable situations and handling emergencies. However, in the production chains of modern knowledge commodities, there remain considerable processes that cannot be completed by humans or machines alone, or through a simple division of labor. It is necessary to continuously open new fields of "human-intelligence collaboration" to restructure traditional production and service sectors such as manufacturing, medical diagnosis, and financial services.

Second, the labor form of human-intelligence collaboration does not negate the subjectivity of humans in labor; human living labor remains the sole source of value creation. In the production logic of generative AI, although intelligent machines possess a certain degree of labor autonomy and can reorganize and drive the development of production, their essence remains the "objectified power of knowledge"—the materialization and extension of human labor. AI itself does not produce new value; rather, by integrating into the social division of labor, it realizes improvements in efficiency and value under human-intelligence collaboration by enhancing labor productivity and expanding the boundaries of labor.

Third, the labor form of human-intelligence collaboration changes the organizational logic of production from the industrial era, driving a shift from labor employment to collaborative cooperation, and from labor organization to collaborative governance. Unlike industrial-era production organizations that fixed people to specific posts, human-intelligence collaborative labor is characterized by the generalization of labor scenes, the high-order nature of labor skills, and the digitalization of the labor process. Traditional "identity recognition" based on labor has turned into system permissions and encrypted keys. This makes data the core bond connecting humans and machines to achieve collaboration. The "downward transmission of decisions" in traditional production hierarchies has turned into a flattened network of "intelligent upward aggregation," allowing laborers to simultaneously implement different types of labor acts across various digital production fields.

Potential Risks of the Penetration of the Logic of Capital. Intelligent large-model production, as a new form of machinery, remains in essence the contemporary manifestation of fixed capital. Its research, development, promotion, and use all require powerful technical and financial support. Therefore, although "human-intelligence collaboration" promotes a natural technical inclusivity and seems to achieve a high-order "composition" of labor products and laborers—benefiting every new type of laborer—the leap in labor power remains subject to the goal of capital valorization. This causes the "living labor of the human to be transformed into a mere living accessory of this machine system," giving rise to a series of intertwined systemic risks such as the imbalance of value distribution, adjustment of employment structures, and the absence of rights protections.

First, the increase in labor productivity and growth in value returns brought by "human-intelligence collaboration" are mostly captured by capital and technical elites, exacerbating the gap between rich and poor in the sphere of labor. Under the monopoly of intelligent technology, the multiplied value returns brought by human-intelligence collaborative labor are seized by intelligent capital, while laborers receive only basic remuneration. This forms a distribution pattern of "technological empowerment, labor weakness," leading to a serious decoupling of value creation from value distribution, and trapping laborers in a new existential predicament of "intelligent poverty" and "data poverty."

Second, "human-intelligence collaboration" further advances labor displacement and the "deskilling" of laborers, leading to waves of "technological unemployment" and structural imbalances in employment. With the support of generative AI, the production capacity of machines has been reinforced to an unprecedented degree. On one hand, the replacement of human labor has deepened from physical strength into intelligence and creativity; on the other hand, the instantaneous generation of knowledge blocks the developmental pathways of complex human cognitive abilities, leading to "deskilling" in a genetic sense. Consequently, "human-intelligence collaboration" drives the employment structure toward "high-order and intelligent" transformation. This not only allows the "quasi-subject" capabilities of AI to replace a large number of repetitive, low-skilled jobs but also creates the structural contradiction of "a shortage of high-end employment and a surplus of low-end employment."

Third, "human-intelligence collaboration" labor conceals the dominance of capital over the labor process and the existing capital-labor relationship, resulting in a lack of protection for laborers' rights. A basic fact that "human-intelligence collaboration" has not changed is that capital's command over labor has not fundamentally shifted; flexible, free, and diversified labor methods are merely masks for this dominance. Laborers are not only caught in the "invisible shackles" of precise algorithmic calculation and strict assessment but also face issues such as the absence of social security, blurred labor boundaries, and obstructed channels for rights protection under the veil of intelligent and flexible labor relations. This makes "human-intelligence collaboration" a form of "self-sacrificing, self-tormenting labor" in the intelligent era that submits to the will of capital.

Innovation of Institutional Logic and Risk Prevention. Preventing and resolving the systemic risks of "human-intelligence collaborative" labor does not mean negating the significance of intelligent technology in transforming labor forms. Rather, it requires using institutional innovation to regulate the logic of capital and master intelligent technology, achieving a dynamic balance between the development of intelligent technology and the relations of production. We must drive "human-intelligence collaborative" labor toward "collaborative evolution" that empowers human development and achieves value sharing.

First, based on the logic of human development and the regulatory system of the socialist system, we must curb the disorderly expansion and value exploitation of AI capital. The developmental logic of capital is the infinite craze for surplus value; as Marx pointed out: "Accumulate, accumulate! That is Moses and the prophets!" Therefore, to unleash the powerful productive forces of "human-intelligence collaboration," we must transcend the logic of infinitely expanding capital and achieve a high-order return to a people-centered logic. This requires giving play to the "civilizing aspect" of capital, allowing it to drive continuous innovation in intelligent technology, while also improving the supervision and regulation of intelligent capital. We must break the monopoly of capital and technical elites over technology, data, and markets, and ensure laborers receive reasonable remuneration through "distribution according to data contribution" and the sharing of intelligent value.

Second, we must master the logic of intelligent technology, ensuring it returns to its essence of serving labor and empowering human development within a governance logic of "AI for good" (智能向善). The key to the development of "human-intelligence collaborative" labor is avoiding the infiltration of the will of a few capital and technical elites into intelligent technology and evading the "hidden rule" of intelligent technology. Therefore, human values, labor rights, and social fairness must be integrated into the entire process of intelligent technology R&D and labor application. We should establish and improve ethical review mechanisms for intelligent technology, increase technical empowerment for low-skilled and vulnerable groups, and promote the inclusive development of intelligent technology.

Third, we must strengthen the cultivation of laborer capabilities, enhancing the use of AI through the improvement of labor skills to drive a revolution in productive forces characterized by mutual human-machine complementarity. Currently, the "human-intelligence collaboration" labor mode has not yet significantly driven the development of productive forces. This is due not only to the knowledge monopoly and technical isolation practiced by capital and technical elites but also to a lack of cultivation of laborers' intelligent application capabilities. Therefore, institutional innovation must incorporate digital literacy and intelligent collaborative capabilities into the education and vocational training systems, elevating the creativity and human-machine collaborative operational skills of laborers to promote the two-way supplementation of human-machine cooperation.

In conclusion, "human-intelligence collaborative" labor in the era of large models heralds a new revolution in productive forces and a future of development. However, it also contains social risks of irrational use. On the basis of analyzing and grasping these risks, we must use institutional innovation to regulate and guide its development, making it a core force for new social progress and human development.