The Labor Theory of Value in the Age of Artificial Intelligence

Marx’s labor theory of value was born in the industrial age, but is it applicable only to the industrial age? This has been a frequently discussed question since the second half of the 20th century. With the proliferation of unmanned production—unmanned workshops, unmanned factories, smart factories—and especially the development of general artificial intelligence, the claim that the labor theory of value is obsolete has once again gained currency. Broadly speaking, the main reason given by those who currently declare Marx’s labor theory of value outdated can be summed up as follows: with the spread of unmanned production, the core idea of the labor theory of value—that “labor is the sole source of value”—has been broken. However, this reason both misunderstands Marx’s labor theory of value and lacks a careful analysis of the development trends of artificial intelligence.

马克思的劳动价值论诞生于工业时代，但是否仅仅适用于工业时代？这是20世纪下半叶以来常被讨论的问题。伴随着无人车间、无人工厂、智能工厂等无人化生产的普及化，尤其是通用人工智能的发展，关于劳动价值论过时了的声音再度流行。大体而言，当前宣告马克思劳动价值理论过时的观点，其主要理由可以归结为伴随无人化生产的普及化，劳动价值论中“劳动是价值的唯一源泉”这一核心思想被打破了。然而，这一理由既误解了马克思的劳动价值论，也缺乏对人工智能发展趋势的细致辨析。

The Emergence of the Collective Worker

总体工人的出现

In Marx, the commodity has a dual character—use-value and value—and this is determined by the dual character of labor. Use-value is produced by concrete labor and is the material bearer of social wealth. Concrete labor differs qualitatively; each type of concrete labor has its corresponding production process. In contrast, in the process of any concrete labor, there is an expenditure of labor-power. This expenditure has no qualitative difference, only a quantitative one, expressed as the length of time labor-power is expended. This process of expenditure is a mere congelation of time; it is abstract labor, corresponding to the value of the commodity, with time as its measure. By means of this distinction between use-value and value, we can see that although artificial intelligence will continuously push laborers out of the sphere of productive labor, giving rise to a large number of unmanned factories, this does not negate Marx’s labor theory of value. This touches on the category of the collective worker that Marx introduced when studying the process of automated production at the time: “the various members of the collective worker either directly or indirectly act upon the object of labor… in order to engage in productive labor, it is no longer necessary to do the work with one’s own hands; it is enough to be an organ of the collective worker and to perform one of its subordinate functions.” [1]1 This passage is a paraphrase of Marx’s concept of the “collective worker” (Gesamtarbeiter) from *Capital*, Volume 1, where he describes how the division of labor within the factory creates a combined collective worker whose members perform different functions, some directly and some indirectly, in the production process.

在马克思那里，商品具有二重性——使用价值和价值，并且是由劳动的二重性决定的。其中，使用价值由具体劳动生产，它是社会财富的物质承担者。具体劳动有质的差别，每种具体劳动都有相应的生产过程。与此不同，在任何一种具体劳动的过程中，都存在劳动力的耗费，这种损耗没有质的差别，而只有量的差别，表现为劳动力耗费的时间长短。这个耗费过程是单纯的时间的凝结，是抽象劳动，它对应的是商品的价值，并以时间为其度量。借助这一使用价值与价值的区分，我们可以看到人工智能虽然会不断把劳动者挤出生产劳动领域，出现大量的无人化工厂，但这并不是对马克思劳动价值论的否定。这里触及马克思当时研究自动化生产过程时所提出的总体工人这一范畴，即“总体工人的各个成员较直接地或者较间接地作用于劳动对象……为了从事生产劳动，现在不一定要亲自动手；只要成为总体工人的一个器官，完成他所属的某一种职能就够了”。

Historically speaking, in the pre-internet era, automated machines were mechanical systems with an internal driving force; they required the supervision and assistance of human workers to complete designated work tasks. At this time, automated machines were not fully automated operations; they always required the presence and assistance of human laborers. In the internet era, automated machines are composed of software and hardware. Humans remotely control their software through the internet as an intermediary, thereby achieving long-distance machine operation. At this point, unmanned factories began to emerge, and human workers and intelligent machines constituted a dualized production scenario mediated by the internet. Entering the era of artificial intelligence, engineers need only design the initial program of the intelligent system’s software; subsequent iterations and updates of the program or algorithm can proceed through its own deep learning and the feeding of massive amounts of data. An integrated collective worker has emerged. Unmanned factories, together with human workers such as programmers, data engineers, and internet users, constitute the various links of the collective worker. The production of value is socialized and holistic, with non-material production or digital production becoming the main mode of social production.

历史来看，在前互联网时代，自动化机器就是有内驱动力的机械系统，它需要在人类工人的监督、协助下才可以完成指定的工作任务。此时，自动化机器并非完全自动化作业，它始终需要人类劳动者的在场与协助。在互联网时代，自动化机器是由软件与硬件构成的，人类通过互联网中介远程操控其软件，从而实现远距离操作机器作业。此时，无人化工厂开始涌现，人类工人与智能机器在互联网媒介下构成了二元化的生产场景。而跨入人工智能时代，工程师只需设计智能系统的软件的初始程序，随后程序或算法的迭代更新，可以通过自身的深度学习与海量数据的投喂而进行，一体化的总体工人出现了。无人化工厂与程序编程师、数据工程师、网民等人类工人一起构成了总体工人的诸环节，价值的生产是社会化的、整体性的，其中，非物质化生产或数字化生产凸显为社会生产的主要方式。

New Issues for the Labor Theory of Value

劳动价值论的新课题

Any kind of labor is a coordination of mental and physical labor, but the proportion of physical and mental labor differs across periods. In the industrial age, humans predominantly exhibited physical labor, with mental labor in a supplementary position; it was not directly transformed into advanced productive forces. In the age of artificial intelligence, mental labor has taken a dominant position, manifested in the labor of intellectuals such as data engineers and programmers. Intellectuals have become the labor force of the AI era. This development trend does indeed challenge the labor theory of value that Marx distilled from material productive labor. Is this challenge the main reason for discarding the labor theory of value? The answer is no. This is not only because Marx’s labor theory of value explicitly discusses the expenditure of mental effort in the production process, but also because mental labor is a form of labor. However, this does not mean we should accept the labor theory of value dogmatically. The correct approach is to inherit and develop Marx’s labor theory of value in light of new historical conditions.

任何一种劳动都是脑力劳动与体力劳动的协同，但体力劳动和脑力劳动的比重在不同时期是不同的。在工业时代，人类更多的是表现出体力劳动，脑力劳动在这里处在辅助性位置，它没有直接转化为先进的生产力。到了人工智能时代，脑力劳动占据了主导地位，它表现为数据工程师、编程师等知识分子的劳动。知识分子成为人工智能时代的劳动力大军。这一发展趋势确实让马克思基于物质生产劳动提炼出来的劳动价值理论受到挑战。这个挑战是构成抛弃劳动价值论的主要理由吗？答案是否定的。这不仅因为马克思的劳动价值论明确论述到生产过程中存在脑力的损耗，也因为脑力劳动是劳动的一种形式。但是，这并不意味着我们要教条式地接受劳动价值论。正确的做法是要结合新的历史条件，对马克思的劳动价值理论予以继承和发展。

Unlike Marx’s era, mental labor currently occupies a dominant position in the social production process. Correspondingly, non-material labor or digital labor is increasingly becoming the main mode of social labor, accompanied by the accelerated emergence of massive amounts of data. Unlike natural objects, data or digital knowledge is the product of human labor; it embodies the labor time humans have invested in it, and therefore includes value. In this regard, the theory of knowledge value proposed and elaborated in detail by Japanese scholar Sakaiya Taichi in his book *The Knowledge-Value Revolution* can be seen as a further development of Marx’s labor theory of value under new conditions, rather than a replacement or falsification. There are at least two reasons for this. First, non-materialized labor is also a form of concrete labor; it does not eliminate the existence of abstract labor. It is just that the mode of production of abstract labor is not necessarily the concrete labor of a specific individual, but is completed at the level of the collective worker through the combination and coordination of multiple work links, including unmanned factories. Second, the “knowledge” in the theory of knowledge value is essentially a product of labor; it does not negate the core idea of the labor theory of value that “labor is the sole source of value.” It is just that, unlike material commodities, knowledge is a non-material commodity, so the carrier of wealth in the current era has also changed. In Sakaiya’s words, “In a ‘knowledge-value society,’ the mainstream of wealth around people will be ‘knowledge-value’ dependent on social subjectivity, while huge sums of money will be abstracted in a form that lacks concrete realizability.” [2]2 Sakaiya Taichi (born 1935) is a Japanese economist and writer. His 1985 book *The Knowledge-Value Revolution* argued that the driving force of the post-industrial economy would shift from material production to the creation of knowledge and information, which he termed “knowledge-value.”

不同于马克思所处的时代，当前脑力劳动在社会生产过程中占据主导地位，与之相适应的是，非物质劳动或数字化劳动日渐成为社会的主要劳动方式，并且海量数据的加速涌现。与自然物不同，数据或数字化知识是人类劳动的产物，它们凝结了人类投入其中的劳动时间，因而包括了价值。就此而言，日本学者堺屋太一在《知识价值革命》一书中提出和详细论证的知识价值论可以视为在新条件基础上对马克思劳动价值理论的进一步发展，而非代替或证伪。理由至少有二：一是非物质化劳动也是一种具体劳动，它并没有消除抽象劳动的存在。只不过抽象劳动的生产方式不一定是具体个人的具体劳动，而是在总体工人这一层面上，以包括无人化工厂在内的多个工作环节的联合、协同来完成的。二是因为知识价值论中的“知识”实质上就是劳动的产物，它并不否定劳动价值论中的“劳动是价值的唯一源泉”这一核心思想。只是与物质性商品不同，知识是非物质性商品，因而当前时代构成财富的载体也发生了变化。用堺屋太一的话说，“在‘知识价值社会’中，人们身边的财富的主流将是依存于社会主观的‘知识价值’，同时，巨额资金则将以不具备具体实现性的形式抽象化”。

However, artificial intelligence is driving a structural transformation in the sphere of productive labor, and the trend towards the subjectification of artificial intelligence poses an unprecedented challenge to Marx’s labor theory of value. On the one hand, this challenge manifests as the distinctions discussed above—the distinction between value and use-value, the distinction between abstract labor and concrete labor, etc.—whose boundaries have become blurred and obscured in the current era of human digital transformation. The theoretical significance of these distinctions in Marx was to grasp the capitalist reality of the industrial age. In the age of artificial intelligence, however, reality includes not only capitalist reality but also socialist reality, and both are undergoing digitization. On the other hand, it manifests as problems arising from the iterative development of artificial intelligence. These problems are particularly prominent around the critical point when AI reaches autonomous consciousness, and are linked to transformations in the order of human labor value and even ethical and moral orders. Below, we will explore this challenge in stages.

然而，人工智能正在推动生产劳动领域的结构性转变，以及人工智能的主体化趋势，使得马克思的劳动价值理论受到了前所未有的挑战。这一挑战，一方面表现为以上所论及的区分，即价值与使用价值的区分，抽象劳动与具体劳动的区分等，其边界在人类遭受数字化转型的当下变得模糊且隐蔽。这些区分的理论意义在马克思那里是要把握工业时代的资本主义现实，而在人工智能时代，现实不仅有资本主义现实，还有社会主义现实，并且都在经历数字化。另一方面，表现为人工智能的迭代发展所引发的问题。这些问题尤其凸显在人工智能达到自主意识的临界点前后，并与人类劳动价值秩序乃至伦理道德秩序的转变关联在一起。下面，我们分阶段探讨这个挑战。

Taking the point at which artificial intelligence possesses autonomous consciousness as a critical threshold, below this threshold AI is, like the automated machinery Marx described, a means of production. The main difference is that the automated machinery Marx described still required human laborers to provide some assistance at the production site, whereas AI, under the influence of the Internet, big data, algorithms, cloud computing, and the like, can carry out unmanned production. The "unmanned" here refers primarily to the absence of human participation at the production site. In other words, it still relies on the directive labor of back-end programmers and/or engineers (which includes, but is not limited to, the design, modification, improvement, and updating of AI programming activities), as well as the online activities of a large number of Internet users (which generate the feedback data necessary for AI optimization and upgrading). From this it follows that the labor of AI is essentially an extension of human labor, and thus it is not the labor of AI that creates value, but human labor. However, because in the age of AI not only the sphere of work but also the sphere of leisure has become digitized, every human individual, whether working or at leisure, participates in the production of raw data. Yet this new phenomenon lies outside the purview of Marx's labor theory of value. Consequently, the concept of the working day, one of the core categories of the labor theory of value, needs to be redefined so that it can incorporate the digital production that takes place during leisure time. Thus, the new working day consists of three line segments, namely a—b—c—d, where the ab segment and the bc segment are the same as Marx's definition, representing necessary labor time and surplus labor time respectively, and the cd segment is newly added; it represents leisure labor time. The digital labor during this period of time creates value in the current era, but its value is either completely ignored, or part of its value is appropriated without compensation by data companies and platform companies.

以人工智能具有自主意识为临界点，在这个临界点之下，人工智能与马克思当年描述的自动化机器一样都是生产工具。差别主要在于，马克思描述的自动化机器依然需要人类劳动者在生产现场施予一定的辅助，而人工智能在互联网、大数据、算法、云计算等的作用下可以进行无人化生产。这里的“无人化”，主要是指生产现场的无人参与。换言之，它依然依赖后台的程序员和/或工程师施予的指令性劳动（它包括但不限于设计、修改、完善、更新人工智能的编程活动），以及大量互联网用户的在线活动（它产生了人工智能优化升级所必需的反馈性数据）。由此而言，人工智能的劳动实质上就是人类劳动的延伸，因而创造价值的不是人工智能的劳动，而是人类的劳动。不过，因为在人工智能时代，不仅工作领域数字化，闲暇领域也数字化，由此人类个体不论工作抑或闲暇，都参与原始数据的生产。但是，这一新现象在马克思的劳动价值论视域之外。由此，作为劳动价值理论之核心范畴之一的工作日概念，需要重新定义，以便能够把闲暇时间的数字化生产吸纳进入。如此，新的工作日，由三条线段构成，即a—b—c—d，其中，ab线、bc线与马克思的定义相同，分别表示必要劳动时间、剩余劳动时间，cd线是新增的，它表示闲暇劳动时间，这段时间的数字劳动在当前时代是产生价值的，但它的价值或被完全忽视，或它的部分价值被数据公司、平台公司无偿占有了。

Above this critical threshold, AI possesses autonomy, attains a moral status equal to that of humans, and becomes another subject alongside humanity. At this point, the labor of the AI subject, like human labor, becomes another source of value. At this stage, the core tenet of Marx's labor theory of value, "labor is the sole source of value," is challenged. Responding to this challenge constitutes another important task for the labor theory of value. There are currently at least two ways to respond to this challenge. The first is to abandon the core tenet of Marx's labor theory of value that "labor is the sole source of value" and to propose a new theory of value on the basis of new realities. This approach is not necessarily Marxist. The second is to accept the core tenet that "labor is the sole source of value" as a premise, but to provide a multi-subject interpretation of the "laborer" who performs this "labor," thereby accommodating the labor of the AI subject. This approach is Marxist or compatible with Marxism.

在这个临界点之上，人工智能具有自主性，达到与人类同等的道德地位，成为与人类并列的另一个主体。在这一点上，人工智能主体的劳动，与人类的劳动一样，成为价值的另一个源泉。在这个阶段，马克思劳动价值论中的这一核心思想“劳动是价值的唯一源泉”就受到挑战。回应这一挑战，构成了劳动价值论的另一重要课题。回应这个挑战的方式，目前来看，至少有两种：一是放弃马克思劳动价值论中“劳动是价值的唯一源泉”这一核心思想，并且在新的现实基础上提出新的价值学说。这种方式不一定是马克思主义的。二是在接受“劳动是价值的唯一源泉”这一核心思想的前提下，对其中的“劳动”的施行者做出多主体的解释，由此兼容人工智能这一主体的劳动。这种方式是马克思主义的或与马克思主义兼容的。

(The author is a professor at the School of Marxism, Hainan University.)

（作者系海南大学马克思主义学院教授）

Online editor: Huihui

网络编辑：慧慧