The Scope and Boundaries of Intelligent Labor

With the rapid development of technologies such as big data, cloud computing, and deep learning, machines possessing human-like intelligence have begun to emerge and enter human life. Intelligent machines constitute a system including intelligent algorithms, data, and robotic arms. By virtue of their efficient computation and information processing, precise operation, and continuous functioning, their widespread application has given rise to phenomena such as “unmanned factories,” “unmanned workshops,” and “unmanned checkouts,” where human living labor is often no longer required on site. Moreover, intelligent machines possess functions such as speech recognition, image processing, editing and translation, and information tracking, enabling them to perform “intelligent” labor encompassing operations like collection, classification, sorting, editing, and positioning, thereby rising from external tools to quasi-subjects or quasi-agents. This means that humans are no longer the sole labor subjects, and the relationship between human society and humans themselves has undergone unprecedented changes. Artificial intelligence not only causes “technological mass unemployment” but also encroaches upon the status of humans as laborers. Consequently, many pessimistically believe that AI labor will replace human labor. However, others argue that, historically, the forms of human labor often undergo stage-based changes with technological progress. In the era of artificial intelligence, although many traditional low-skill occupations will gradually disappear, human labor will increasingly shift toward innovative fields and will not be completely replaced. So, what exactly is the scope and boundary of intelligent labor replacing human labor?

随着大数据、云计算、深度学习等技术的迅猛发展，具有类人智慧的机器开始出现并进入人类生活。智能机器是包括智能算法、数据与机械臂在内的智能机器体系，凭借高效运算与信息处理、精准操作、持续运转等，它的广泛应用造成大量“无人工厂”“无人车间”“无人柜台”等现象，人类的活劳动很多时候不必在场。不仅如此，智能机器拥有语音识别、图像处理、编辑翻译、信息追踪等功能，能够进行涵盖搜集、分类、整理、编辑、定位等运作方式的“智能化”劳动，从以往的外在工具上升为类主体或拟主体。这意味着人不再是单一劳动主体，人类社会与人本身的关系发生空前变化。人工智能不仅造成“技术性广泛失业”，还僭越了人类的劳动者身份。于是，许多人悲观地认为，人工智能劳动将替代人类劳动。不过，也有人认为，从历史上看，人类的劳动形态往往随着技术进步呈现阶段性变化。人工智能时代，虽然许多技术含量低的传统职业会逐渐消失，但是人类劳动将更多向创新领域转移，不会被完全取代。那么，智能劳动替代人类劳动的范围与边界究竟在哪里？

With the advancement of science and technology, intelligent devices created by artificial intelligence—such as automated teller machines, unmanned logistics warehouses, vending machines, securities analysis software, intelligent editors, automated ticket machines, and intelligent inspection equipment—have partially replaced the labor of workers in logistics, sales, securities, journalism, and other fields. Although some argue that this kind of replacement labor is merely labor carried out with the help of production factors such as human thought, data information, and material machines, and is therefore only a transfer of information and intelligence, fundamentally incapable of generating value or engaging in exchange on its own, here intelligent machines and humans still “manifest as links in the system of social labor.” [1]1 Karl Marx, *Grundrisse der Kritik der Politischen Ökonomie* (1857–1858), in *Marx-Engels-Werke* (MEW), vol. 42, p. 601. In terms of labor characteristics, the labor that can be replaced by intelligent machines all exhibits repetitive, patterned, and procedural features. Such labor does not require humans to invest excessive emotion, will, or experience, nor does it necessitate constant change and renewal. Humans can materialize production factors such as knowledge and information into intelligent machines, thereby transforming human living labor into materialized labor, and thus replacing humans in performing part of their mental or physical labor.

随着科学技术的进步，人工智能造就的自动柜员机、无人物流仓库、自动售货机、证券分析软件、智能编辑、自动售票机、智能检查设备等智能设备，部分取代了物流、销售、证券、新闻等方面从业人员的劳动。尽管有人认为这种替代劳动不过是借助人类思维、数据信息、物质机器等生产要素开展的劳动，因此只是一种信息和智慧的转移，根本不可能产生价值，自身也无法进行交换，但是在这里智能机器与人类之间仍然“表现为社会劳动体系的各个环节”。从劳动特质上看，能够被智能机器取代的劳动都具有重复性、模式化、程序化特点。它们不需要人类投入过多情感、意志、经验，也没有必要不断变更和翻新。人类可以通过把知识、信息等生产要素物化到智能机器身上，实现人的活劳动向物化劳动的转化，从而代替人类完成部分脑力劳动或体力劳动。

Marx pointed out: “The path of the machine system is decomposition—achieved through the division of labor, which gradually turns the worker’s operations into mechanical operations.” [2]2 Karl Marx, *Das Kapital*, vol. 1, in *Marx-Engels-Werke* (MEW), vol. 23, p. 442. Specifically, artificial intelligence, as a system of intelligent machines including intelligent algorithms, data, and robotic arms, also achieves the replacement of the human brain and muscles through functional decomposition. First, algorithms are the brain of AI, capable of replacing human computational ability and basic cognitive ability. Algorithms are the key element of AI, formed by engineers through modeling based on mining, selecting, and processing data according to certain logic, frameworks, and procedures. Currently, the mainstream AI algorithm is the neural network algorithm, which has achieved transformative development from shallow learning to deep learning. Neural network algorithms preset more model parameters and, by combining low-level features to form more abstract high-level attribute categories or features, discover distributed feature representations of data, enabling intelligent machines to have analytical learning abilities like humans, capable of recognizing data such as text, images, and sounds, thereby replacing human mental labor. Second, large amounts of data can depict human experience. AI processes source data through collection, labeling, and other steps to form corresponding datasets. Subsequently, after extensive training, a well-performing model is generated that covers as many scenarios as possible. On this basis, by finding patterns in the data, predictions and decisions are made for specific tasks, carrying out “human-like” work. Third, the robotic arm is the end effector of AI, including various forms such as mechanical grippers and mechanical hammers. Robotic arms, by grasping, moving objects, or operating automatic devices according to fixed programs, achieve mechanization and automation of production, and can replace humans in performing repetitive, patterned, and procedural labor.

马克思指出：“机器体系的这种道路是分解——通过分工来实现，这种分工把工人的操作逐渐变成机械的操作。”具体来说，人工智能作为包括智能算法、数据与机械臂在内的智能机器体系，也是通过功能分解实现对人脑、肌肉的替代。第一，算法是人工智能的大脑，可以替代人类的计算能力和基础认知能力。算法是人工智能的关键要素，它由工程师根据一定的逻辑、框架、规程，在对数据进行挖掘、选择、处理的基础上建模形成。目前，主流的人工智能算法是神经网络算法，实现了从浅层学习向深度学习的变革式发展。神经网络算法预设了更多模型参数，通过组合低层特征形成更加抽象的高层属性类别或特征来发现数据的分布式特征表示，让智能机器像人类一样具有分析学习能力，能够识别文字、图像和声音等数据，实现对人类脑力劳动的替代。第二，通过大量数据可以刻画人类经验。人工智能对源数据进行采集、标注等处理后形成相应数据集。之后，经过大量训练生成一个良好模型，覆盖尽可能多的场景。在此基础上，通过找寻数据中的规律对具体任务进行预测和决断，开展“类人”工作。第三，机械臂是人工智能的末端执行器，包括机械爪、机械锤等多种形态。机械臂通过按固定程序抓取、搬运物件或操作自动装置，实现生产的机械化和自动化，可以代替人类从事重复性、模式化、程序化劳动。

With the rapid development of AI technology, some believe that intelligent labor will replace human labor, and thus Marxist theories on labor have become invalid. This is not the case! As long as we apply the Marxist materialist conception of history and theory of labor to analyze the characteristics and value of AI, we can break through the surface of intelligent labor and clearly see its boundaries and limitations.

随着人工智能技术的快速发展，有人认为智能劳动将会取代人类劳动，因此马克思主义有关劳动的理论失灵了。其实不然！只要运用马克思主义唯物史观和劳动观分析人工智能的特征、价值，就能突破智能劳动的表象，看清楚智能劳动的边界和局限性。

First, intelligent labor does not spontaneously achieve “creation from nothing”; rather, it relies on the large amounts of data and information fed by humans. AI is an intelligent device forged by humans using engineering methods and simulation methods to realize the functions of human limbs and brains. Without external empowerment, it cannot function. As a carrier of scientific transformation and technological undertaking, AI is essentially a programmed and standardized “consciousness.” Although it possesses “human-like” abilities in the production process, elevating itself from a manipulated object to a limited executor, this autonomy is very limited, no different from production tools like hand mills or steam mills, and fundamentally incapable of achieving “creation from nothing” or reflecting the initiative of mental labor. In contrast, humans do not need much data, or can make predictions and decisions with very little data; true geniuses may not need any data at all, and through imagination and innovation, can transform all things in the world into “natural materials suitable for human needs,” [3]3 Karl Marx, *Das Kapital*, vol. 1, in *Marx-Engels-Werke* (MEW), vol. 23, p. 192. realizing the process of creation from nothing.

第一，智能劳动不会自发实现“从无到有”的生产，而是依赖人类投喂的大量数据和信息。人工智能是人类通过运用工程学方法与模拟方法等锻造的一个实现人类四肢和头脑功能的智能设备，离开外在赋能，它将寸步难行。作为科学转化与技术承接的载体，人工智能本质上是程序化、标准化的“意识”，虽然在生产过程中具有“类人”本领，使自身从被操纵者一跃成为有限的执行者，但这种自主性非常有限，无异于手推磨、蒸汽磨等生产工具，根本不可能实现“从无到有”的生产，也无法体现脑力劳动的能动性。与之相比，人类不需要太多数据，或者仅仅掌握极少数据就能做出预测和决断；真正的天才甚至不需要什么数据，通过想象能力、创新能力就可以使世间万物变为“适合人的需要的自然物质”，实现从无到有的过程。

Second, AI does not fundamentally create value; the sole source of value remains human living labor. Intelligent labor is not machines autonomously and spontaneously carrying out labor, but rather “a new form of living labor that humans endow themselves with through technological forms.” [4]4 This is a paraphrase of a common formulation in Chinese Marxist political economy discussions of AI. Although intelligent labor is direct labor acting on commodity production, it can never detach from humans as the indirect labor subject. According to the Marxist theory of labor value, the value condensed by labor can only be generated in commodity production activities and manifested through exchange relations between people. Therefore, the sole source of value remains and can only be human living labor. If AI were considered the source of value, that would commit the error of vulgar bourgeois economics. No matter how sophisticated and efficient intelligent machines become, they are merely “a measure of the development of human labor power” and “an indicator of social relations,” [5]5 Karl Marx, *Das Kapital*, vol. 1, in *Marx-Engels-Werke* (MEW), vol. 23, p. 194. fundamentally still belonging to the category of labor intermediaries, and utterly incapable of replacing human living labor.

第二，人工智能本质上不创造价值，价值的唯一源泉仍然是人的活劳动。智能劳动不是机器自主自在地开展劳动，而是“人们以技术形式赋予自身活劳动的新形态”。虽然智能劳动是作用于商品生产的直接劳动，但它永远无法脱离作为间接劳动主体的人类。依据马克思主义劳动价值观，劳动凝结的价值只能生成于商品生产活动中并通过人与人的交换关系表现出来。因此，价值的唯一源泉仍然是而且只能是人类的活劳动。如果说人工智能是价值的来源，那就犯了资产阶级庸俗经济学错误。无论智能机器发展到多么精巧和高效，也只是“人类劳动力发展的测量器”和“社会关系的指示器”，其本质上仍属于劳动中介范畴，根本无法取代人类的活劳动。

Third, AI lacks human qualities such as emotion, will, and aesthetic judgment, and therefore lacks adaptability in the production process and uniqueness in production outcomes. Intelligent machines, after all, belong to the material world and cannot possess a conscious world; thus, AI has intelligence but no wisdom. As Kant said, “Science may be invincible in cognition, but it is powerless in the face of emotion and will.” [6]6 Immanuel Kant, *Critique of Judgment* (1790), Introduction, Section II. The exact wording is a paraphrase; Kant’s original distinguishes between the faculties of cognition, feeling of pleasure/displeasure, and desire. Intelligent labor, as unconscious instrumentalized labor, can neither make causal judgments nor analyze connections between things, let alone grasp the essence of things. Consequently, when faced with unexpected situations or the need to update production methods, it cannot make emergency responses or change processing strategies. Additionally, AI programs are essentially passive and mechanical physical responses to the objective world; their operational processes are repetitive and procedural, so the labor process and outcomes are highly consistent, lacking the creativity and specificity of mental labor.

第三，人工智能不具备人类的情感、意志、审美等品格，因而生产过程中缺乏应变能力、生产效果上缺少独特性。智能机器毕竟属于物质世界范畴，不可能拥有意识世界，因此人工智能有智力没智慧。正如康德所言，“科学可能在认知方面所向无敌，但面对情感、意志则无能为力”。智能劳动作为无意识的工具化劳动，既无法进行因果关联判断，分析事物之间的联系，更无法把握事物本质，进而在面对意外状况或需要更新生产方法时，无法作出应急反应或转变处理策略。另外，人工智能程序本质上是对客观世界作出的消极被动物理性反应，其运作流程是重复性和程序化的，因而劳动过程和劳动结果具有高度一致性，不具备脑力劳动的创造性和特殊性。

In summary, although AI has achieved significant breakthroughs in areas such as human genome sequencing and the structural decoding of the human brain, constrained by the boundaries and limitations of AI technology, it can never shed its tool-like characteristics, nor can it change its own “belonging to matter” attribute. However, as an intelligent tool that partially replaces human labor, AI has already surpassed manual tools and mechanical tools, representing a great historical advance; while it creates new forms of labor alienation in the process of development, it also holds positive significance for realizing the liberation of human labor.

总之，尽管人工智能在人类基因组排序、人类大脑结构性解密等方面都取得了重大突破，但是受人工智能技术的边界和局限性制约，它永远摆脱不了工具特征，也永远改变不了自身的“属物”属性。但是，人工智能作为一种有限替代人类劳动的智能工具，已经超越了手工工具和机械工具，是历史的一大进步；它在发展过程中造成新型劳动异化现象的同时，也对实现人类劳动解放具有积极意义。

Online editor: Paul

网络编辑：保罗