Generative Artificial Intelligence and New Horizons in the Value Movement of Digital Capitalism

In recent years, the high innovation, strong penetration, and broad coverage of the new generation of Artificial Intelligence (AI) technology have exerted extensive and profound impacts on the value production, realization, and distribution of digital capitalism. In particular, the emergence and dazzling iterative upgrades of generative artificial intelligence represented by ChatGPT have greatly stimulated the value potential of data elements and promoted an overall leap in digital productive forces, marking the entry of human knowledge production into the era of large models. The history of capitalist development shows that every major technological innovation and its derivative effects trigger adjustments in relations of production and transformations in the production system. At present, the production, exchange, circulation, and distribution of digital capitalism are accelerating toward intelligence, and “artificial intelligence capitalism” has begun to take shape.

近年来，新一代人工智能（Artificial Intelligence，AI）技术的高创新、强渗透和广覆盖，已经对数字资本主义价值生产、实现和分配产生广泛而深刻的影响。特别是以ChatGPT为代表的生成式人工智能的横空出世和令人眼花缭乱的升级迭代，极大地激发了数据要素的价值潜能，并促进了数字生产力的整体跃升，标志着人类知识生产进入了大模型时代。资本主义发展史表明，每一次重大技术创新及其衍生效应都会引发生产关系的调整和生产体制的变革。目前来看，数字资本主义的生产、交换、流通和分配加速智能化，“人工智能资本主义”已初显端倪。

From the perspective of Marxist political economy, the basic trajectory of the entanglement between capital and labor, technology and institutions under the capitalist mode of production is that capital strives to use technological progress to discipline labor in order to extract surplus value. At the same time, however, the “subjectification” of dead labor [1]1 Karl Marx, *Capital*, Vol. 1, Chapter 26. under the enhancement of technology inevitably backfires on value itself. Faced with the phenomenal craze triggered by ChatGPT’s “breaking out of the circle,” [2]2 Kai-Fu Lee, *AI Superpowers: China, Silicon Valley, and the New World Order* (Houghton Mifflin Harcourt, 2018), p. 112. a pressing issue worthy of in-depth exploration is: in what “innovative” ways will the technological emergence of generative artificial intelligence and the comprehensive upgrade of intelligent machines enact the value movement of digital capitalism?

在马克思主义政治经济学的视域中，资本主义生产方式下资本与劳动、技术与制度相互纠缠的基本进路就是资本竭力利用技术进步来规训劳动以攫取剩余价值，但同时，技术加持下死劳动“主体化”也不可避免地反噬价值。面对ChatGPT“出圈”引发的现象级热潮，一个亟待关注且值得深入探讨的问题就是生成式人工智能的技术出场及智能机器的全面升级，会以何种“创新”方式来演绎数字资本主义价值运动？

To deepen the understanding of the developmental laws of digital capitalism, it is necessary to conduct a preliminary exploration of the following frontier issues. First, from the perspective of commodity production in digital capitalism, generative artificial intelligence is unlocking a new cognitive revolution—what new changes will this bring to the production and consumption of digital commodities? Second, from the perspective of the labor process in digital capitalism, what new characteristics emerge in the value creation of digital labor under human-machine collaboration? Third, from the perspective of capital accumulation in digital capitalism, what new trends will the competition among digital oligarchs over large models lead to in digital capital accumulation? Furthermore, with the accelerated development of artificial intelligence technology, what new transformations will the production order of digital capitalism face?

为深化对数字资本主义发展规律的认识，有必要对如下前沿议题开展初步探讨。一是从数字资本主义商品生产来看，生成式人工智能正在解锁新的认知革命，这会给数字商品产消带来哪些新变化？二是从数字资本主义劳动过程来看，人机协作下的数字劳动价值创造呈现出哪些新特征？三是从数字资本主义资本积累来看，数字寡头的大模型竞争将会导致数字资本积累呈现哪些新趋势？此外，随着人工智能技术的加速发展，数字资本主义生产秩序将面临哪些新的变化？

I. Cognitive Revolution and a New Stage in the Production and Consumption of Digital Commodities

一、认知革命与数字商品产消的新阶段

“The Taylorism of the 20th century sparked a ‘productivity revolution’ that rapidly increased labor productivity. In digital capitalist society, how to initiate a productivity revolution in the realm of knowledge—that is, a ‘knowledge productivity revolution’—will become the key to development.” [3]3 Karl Marx, *Capital*, Vol. 1, Chapter 15. At present, the rapid advance of generative artificial intelligence technology is initiating an industrial revolution in the field of human knowledge production. The automation of information and knowledge production under capitalism is gradually becoming a reality, thereby breaking through the capacity constraints of digital commodity production and accelerating the entry into a new stage of hyper-personalized customization.

“20世纪的泰罗制掀起了急速提高劳动生产率的‘生产率革命’，而在数字资本主义社会，如何掀起知识版的生产率革命，也就是‘知识生产率革命’，将成为发展的关键。”当下，生成式人工智能技术狂飙突进，正在开启人类社会知识生产领域的工业革命。资本主义信息和知识生产自动化逐渐成为现实，从而使数字商品生产突破产能限制，加速迈进超个性化定制新阶段。

(I) “Synthetic Data” Breaks the Supply Constraints on Data Elements

（一）“合成数据”打破数据要素供给约束

Entering the era of the digital economy, data has become a key production factor. In practice, massive, heterogeneous data, after factorized processing such as manual cleaning, becomes the “fuel” for intelligent machines, the “raw material” for digital commodities, and the “asset” of digital platforms. However, in recent years, the commercial practice of analytical artificial intelligence has shown that the production and consumption of digital commodities face increasingly tight supply constraints on digital elements. The reasons are as follows: existing data resources such as text, images, and audio vary widely in quality and struggle to cover long-tail and edge-case data, resulting in insufficient data diversity; the collection and labeling of AI training data are time-consuming, labor-intensive, and costly; factors such as government data regulation, user privacy protection, and digital geopolitics increase the uncertainty of data element supply; and the deepening segmentation of data commodity markets highlights the “shortage” of proprietary data in vertical domains, among other issues.

进入数字经济时代，数据成为关键的生产要素。在实践中，海量、异构数据经过人工清洗等要素化处理后，成为智能机器的“燃料”、数字商品的“原料”和数字平台的“资产”。但是，近年来分析式人工智能的商业化实践表明，数字商品产消面临日渐趋紧的数字要素供给约束。原因在于，已有的文本、图像和音频等数据资源质量良莠不齐，难以覆盖长尾、边缘案例数据，数据多样化不足；人工智能训练数据采集和标注费时费力、成本高企；受政府数据规制、用户隐私保护和数字地缘政治等因素的影响，数据要素供给的不确定性上升；数据商品市场细分加深，垂直领域专有数据“短缺”现象凸显，等等。

Currently, AI large model technology is creating new momentum for data valorization. The ultimate pursuit and stunning performance of generative artificial intelligence in “human-like intelligence” are inseparable from the data feeding of meticulously organized, high-quality corpora. According to reports, the AI language model GPT-3 was trained using 45 terabytes (TB) of data, equivalent to approximately 450,000 years of text from the People’s Daily. [4]4 See, for example, Nick Srnicek and Alex Williams, *Inventing the Future: Postcapitalism and a World Without Work* (Verso, 2015), p. 171. The dialogue programming language model LaMDA’s training data came from 1.56 trillion words scraped from the internet. The contrastive language-image pre-training model CLIP collected over 400 million “text-image” training data pairs from the internet. For digital capital, “data hunger” is fiercer than ever before. In response, digital tech giants are no longer satisfied with “data enclosure” and “data colonization”; instead, they are exploring new paths by actively developing synthetic data technologies to efficiently achieve the large-scale production of data elements through data augmentation, simulated generation, and simulation environment data collection.

当下，AI大模型技术正在打造数据价值化新动能。生成式人工智能对“类人智能”的极限追求和惊艳表现，须臾离不开精心组织的高质量语料库的数据投喂。据报道，人工智能语言模型（GPT-3）使用45太字节（TB）数据量进行训练，约为《人民日报》45万年的文字量。对话编程语言模型（LaMDA）的训练数据来自从互联网抓取的1.56万亿个词汇。对比语言—图像预训练模型（CLIP）搜集了互联网上超过4亿对“文本—图像”训练数据。对于数字资本而言，“数据饥渴”比以往任何时候都来得更加猛烈。对此，数字科技巨头不再满足于“数据圈地”和“数据殖民”，而是另辟蹊径，积极开发合成数据技术，通过数据增强、模拟生成和仿真环境数据采集来高效实现数据要素的规模化生产。

Synthetic data can mathematically or statistically reflect the properties of real-world data and can thus serve as a substitute for real-world data to train, test, and validate AI large models. Moreover, the cost of synthetic data is far lower than that of real data. Some institutions predict that 60% of the data used to train large models in 2024 will be synthetic data, and it is expected that by 2030, the vast majority of data used by AI large models will be synthetically generated by AI. In this way, synthetic data created based on AI content generation technology will become prevalent. It can not only automatically generate safe, labeled, pre-processed standard data to satisfy the increasingly “hungry” AI large models, but the automation of data element supply also allows digital capital to further shake off the “shackles” of digital labor, more wantonly driving intelligent machines forward at full speed to achieve the ultimate harvesting of a new round of “data dividends” from the intelligent economy.

合成数据可以在数学或统计学上反映真实世界数据的属性，因而可以作为真实世界数据的替代品来训练、测试、验证AI大模型。并且，合成数据的成本要远远低于真实数据。有机构预测，2024年用于训练大模型的数据中有 60%将是合成数据，预计2030年AI大模型使用的绝大部分数据都将由人工智能合成。如此一来，基于人工智能内容生成技术创建的合成数据大行其道，不仅可以自动生成具有安全性、标签化、经过预处理的标准数据来满足越来越“饥渴”的AI大模型，而且数据要素供给的自动化，使数字资本进一步摆脱数字劳动的“羁绊”，更加恣意地驾驭智能机器一路狂奔，以实现对智能经济新一轮“数据红利”的极限收割。

(II) Model Pre-training Promotes the “Generalization” of Intelligent Machine Productive Forces

（二）模型预训练推动智能机器生产力出现“泛化”

Under the capitalist mode of production, “the instruments of labor go through various morphological changes, the final form of which is the machine, or more precisely, an automatic system of machinery.” [5]5 Karl Marx and Friedrich Engels, *The Communist Manifesto*, Chapter 1. Undoubtedly, in the stage of digital capitalism, the iterative upgrading of intelligent machines is, with a breathtaking technological acceleration, fully demonstrating the “autonomy” of the “system of machinery.” However, to the “regret” of digital capital, in the application scenarios of analytical artificial intelligence, intelligent machines appearing in concrete forms such as search engines, facial recognition, intelligent voice, and algorithmic recommendations are all focused on a single modality and specialized in pattern recognition and behavior prediction within specific domains. Borrowing a term from new institutional economics, intelligent machines exhibit strong “asset specificity”: “what is taught may be learned, but what is not taught will not be learned.” The “model island” phenomenon greatly constrains the digital production potential of intelligent machines.

在资本主义生产方式下，“劳动资料经历了各种不同的形态变化，它的最后的形态是机器，或者更确切些说，是自动的机器体系”。毋庸置疑，进入数字资本主义阶段，智能机器的升级迭代正在以令人窒息的技术加速度将“机器体系”的“自主性”演绎得淋漓尽致。但是，令数字资本“抱憾”的是，在分析式人工智能的应用场景中，以搜索引擎、人脸识别、智能语音和算法推荐等具化形态出现的智能机器都是专注于单一模态、专事于特定领域的模式识别和行为预测。借用新制度经济学的术语，智能机器具有强“资产专用性”，“教过的可能会，没教过的就不会”，“模型孤岛”现象极大地制约了智能机器的数字生产潜力。

In contrast, under the engineering combination of “big data + large computing power + strong algorithms,” AI large models can achieve “what is not taught is also learned” through “brute force creating miracles.” [6]6 Hartmut Rosa, *Social Acceleration: A New Theory of Modernity* (Columbia University Press, 2013), p. 71. From the perspective of manufacturing mechanisms, digital capital adopts advanced neural network architectures, “pre-training” foundational models by “devouring” massive unlabeled datasets, and then fine-tuning the models based on reinforcement learning from human feedback (RLHF). With only a small number of samples or even zero-shot learning, they can successfully complete new tasks and meet specific purposes. That is, AI large models possess significant downstream task adaptability, promoting the “generalization” of intelligent machine productive forces and enabling “industrial assembly-line deployment across different industries, vertical domains, and functional scenarios.” [7]7 Karl Marx, *Grundrisse*, Notebook VII, “The Fragment on Machines.” In other words, under the enhancement of generative artificial intelligence technology, the standardized development and ubiquitous application of intelligent machines can produce a dazzling “intelligence emergence” effect, thereby providing digital capitalism with virtual production lines installed on cloud-network terminals for the automated production of digital commodities. Once the production of intelligent machines by intelligent machines becomes a reality, “artificial intelligence capitalism” will acquire its own specific means of production and thereby establish a technological foundation suited to itself.

相比之下，在“大数据+大算力+强算法”工程性结合下，AI大模型可以“大力出奇迹”地实现“没教过的也会”。从制造机理来看，数字资本采用先进神经网络架构，通过“吞噬”巨量无标注数据集来“预训练”基础模型，然后基于人类反馈强化学习（RLHF）进行模型精调，只需小样本甚至零样本学习，便能成功地完成新的任务，满足特定的用途。也即，AI大模型具有显著的下游任务适配能力，推动智能机器生产力“泛化”，实现“在不同行业、垂直领域、功能场景的工业流水式部署”。换言之，在生成式人工智能技术的加持下，智能机器的标准化开发和泛在化应用可以形成令人应接不暇的“智能涌现”效应，从而为资本主义数字商品自动化生产提供装置于云网端的虚拟生产线。一旦用智能机器生产智能机器成为现实，“人工智能资本主义”就将掌握它特有的生产资料，进而建立起与自己相适应的技术基础。

(III) AI Content Generation Technology Promotes the Automation of Digital Content Production

（三）人工智能内容生成技术推动数字内容生产自动化

Entering the stage of digital capitalism, digital content production has become the use-value production most characteristic of the era in capitalist society. From professionally generated content (PGC) in the Web 1.0 era, to user-generated content (UGC) in the Web 2.0 era, and now leaping to AI-generated content (AIGC) in the present, the development and application of digital technology continuously drive changes in the mode of digital content production.

进入数字资本主义阶段，数字内容生产成为了资本主义社会最具时代特征的使用价值生产。从Web1.0时代专业生成内容（PGC）发展到Web2.0时代用户生成内容（UGC），再跃迁到当下人工智能生成内容（AIGC），数字技术的发展与应用持续推动着数字内容生产模式的变革。

As shown in Table 1, from the perspective of content production methods, whether it is text generation, script creation, or webpage translation, site building, and code writing, AI-generated content both overcomes the shortcoming of “slow output” in professionally generated content and resolves the drawback of “a mix of good and bad” in user-generated content, becoming the “factory” and “assembly line” for automated digital content production. This production method can ultimately satisfy massive personalized demands in a low-marginal-cost, high-efficiency manner. Precisely for this reason, with its rapid response capability, vivid knowledge output, and rich application scenarios, AI content generation is increasingly becoming a new engine for both material and spiritual production in human society.

如表1所示，从内容生产方式来看，无论是文本生成、脚本绘制，还是网页翻译、站点制作和代码编写，人工智能生成内容既克服了专业生成内容“出货慢”的不足，又消解了用户生成内容“鱼龙混杂”的弊端，成为数字内容自动化生产的“工厂”和“流水线”。这种生产方式最终可以实现以低边际成本、高效率的方式满足海量个性化需求。正因如此，凭借快速的反应能力、生动的知识输出、丰富的应用场景，人工智能内容生成日趋成为人类物质生产和精神生产的新引擎。

Furthermore, AI content generation technology empowers the development of various industries, giving rise to new digital economy products, new business forms, and new models, further consolidating the value system of digital capitalism. First, in commercial applications such as text creation, human-computer interaction, audio-visual media, education, and finance, generative AI products can reach “human-like” or even “superhuman” levels; their “usefulness of things” is astonishing, significantly enhancing user stickiness and willingness to consume. Second, “Model as a Service” (MaaS) realizes the transformation of AI applications from “handicraft workshops” to “factory models.” Under the MaaS business model, digital capital charges based on model invocation volume (i.e., according to the amount of data requests and actual computation), achieving pricing by “data,” which greatly reduces the market transaction costs of digital content commodities.

进一步地，人工智能内容生成技术赋能各个行业的发展，催生数字经济新产品、新业态、新模式，进一步巩固了数字资本主义价值体系。首先，在文字创造、人机交互、影音、教育、金融等场景商业化应用中，生成式人工智能产品可以达到“类人”甚至“超人类”水平，其“物的有用性”令人惊叹，显著增强了用户粘性和用户消费意愿。其次，“模型即服务”（MaaS）实现了人工智能应用从“手工作坊”向“工厂模式”的转变。在“模型即服务”的商业模式下，数字资本基于模型调用量（也即根据数据请求量和实际计算量）收费，实现按“数”定价，极大地降低了数字内容商品的市场交易成本。

Moreover, after the emergence of generative AI technology, digital capital can not only more easily realize the materialization and appropriation of society’s “general intellect” using machine intelligence but also unprecedentedly “infrastructuralize” intelligent machines and firmly control them in its own hands. The result is that digital capital monopolizes the “means of cognition,” further consolidating the ownership foundation of digital commodity production.

再者，生成式人工智能技术涌现后，数字资本不仅可以利用机器智能更加轻松地实现对社会“一般智力”的物化和占有，而且还前所未有地将智能机器“基础设施化”，并牢牢地掌控在自己的手中。其结果是，数字资本独占“认知性生产资料”（means of cognition），进一步夯实了数字商品生产的所有制基础。

(4) “Human-Machine Dialogue” Promotes Upgraded Digital Consumption Experience

（四）“人机对话”促进数字消费体验升级

Since the birth of the computer, the “atomic world” of human material production and daily life has been continuously “bitified,” and increasingly rich digital products and services have profoundly changed the consumption landscape of contemporary capitalist masses. Especially after entering the Web 2.0 era, the easy-to-operate and user-friendly graphical user interface (GUI) greatly dissolved the digital divide brought by “command line” operations; “using machines” and “going online” are no longer geek novelties, and ordinary people can easily enjoy convenient digital services. But fundamentally, analytical AI statically displays information, and its digital content presents an ineradicable algorithmic “coldness” and technological alienation in human-computer interaction. Moreover, algorithmic manipulation such as “big data price discrimination against regular customers” [8]8 “Big data price discrimination against regular customers” (大数据杀熟) refers to the practice where platforms use consumer data to charge higher prices to existing or loyal customers compared to new users. frequently infringes on user rights, further exacerbating user distrust in human-computer interaction.

自计算机诞生以来，人类物质生产生活的“原子世界”不断被“比特化”，日益丰富的数字产品和服务深刻地改变了当代资本主义大众的消费景观。特别是进入Web2.0时代后，操作便捷、用户友好的图形用户界面（GUI）极大地消解了“命令行”操作带来的数字鸿沟，“用机”“上网”不再是极客新潮，普通民众也可以轻松享受便捷的数字服务。但从根本上讲，分析式人工智能静态地展示信息，其数字内容呈现出无法消弭人机交互中的算法“冷漠”和技术疏离感。并且，“大数据杀熟”等算法操纵还频频侵害用户权益，进一步加剧了人机交互中的用户不信任感。

Currently, the “breaking out of the circle” of ChatGPT [9]9 “Breaking out of the circle” (破圈) is a Chinese internet term meaning a phenomenon or product gaining mainstream popularity beyond its original niche community. is accelerating the deployment of generative AI technology and driving an upgrade in the consumption experience of digital commodities. Its minimalist interaction based on natural language instantly immerses ordinary users in a more natural, direct, and smooth conversational digital scene, bringing them a refreshing digital consumption experience. This is mainly reflected in the following aspects: in real-time information retrieval, generative search engines present results in a chapter-reading format, replacing the simple list-style link display of traditional search engines, placing information more quickly and comprehensively at users’ fingertips, greatly catering to ordinary users’ psychology of seeking “instant answers”; in digital service provision, generative AI’s embedded and plug-in integration transforms the interaction methods of massive applications, with functions covering areas such as online ordering, enterprise office work, process optimization, graphic and text processing, product marketing, and recommendations, achieving enhanced functionality of mobile apps. Furthermore, through precise targeting that “understands what you think, knows what you need, and solves what worries you,” and with full participation from intelligent customer service, intelligent marketing, digital human live streaming, etc., it brings an immersive experience of virtual-real integration and multi-sensory interaction to online consumption.

当下，ChatGPT的“破圈”加速了生成式人工智能技术的落地，正在推动数字商品消费的体验升级。它基于自然语言的极简交互，使普通用户瞬间沉浸在更加自然、直接和顺畅的对话式数字场景之中，为其带来的数字消费体验令人耳目一新。这主要体现在以下方面：在实时信息检索上，生成式搜索引擎以篇章阅读的方式呈现，替代了传统搜索引擎简单的摘要罗列式链接显示，更快捷且全面地将信息放在指尖，极大地迎合了普通用户期待“秒答”的求问心理；在数字服务供给上，生成式人工智能嵌入式、插件式融合，变革海量应用的交互方式，其功能覆盖了在线点餐、企业办公、流程优化、图文处理、产品营销和推荐等领域，实现了手机软件功能的增强。进而，以“懂你所想、知你所需、解你所忧”的精准触达，以智能客服、智能营销、数字人直播等全力加盟，给线上消费带来了虚实相融、多感官交互的沉浸式体验。

II. New Characteristics of Human-Machine Collaboration and Digital Labor Value Creation

二、人机协作与数字劳动价值创造的新特征

Entering the digital economy era, analytical AI has initiated capitalism’s “second machine revolution” and also spawned various forms of digital labor. At present, the technological applications of generative AI are springing up like bamboo shoots after rain. “With the support of large models, AI is transforming from a ‘tool’ into a ‘partner’ for humans.” Human-machine collaboration will continuously decompose the tasks of digital labor, fundamentally changing the methods and content of completing digital work, and the value creation of digital labor will generally exhibit new characteristics of “intelligent enhancement.”

进入数字经济时代，分析式人工智能开启了资本主义“第二次机器革命”，也催生了形态各异的数字劳动。目前来看，生成式人工智能的技术应用如雨后春笋般涌现。“在大模型的加持下，人工智能正在从‘工具’变成人类的‘伙伴’。”人机协作将使数字劳动的任务被不断拆解，数字化工作的完成方式和完成内容也将发生根本性改变，数字劳动价值创造总体上呈现“智能增强”的新特征。

(1) Human-Machine Collaboration Promotes the “Value Elevation” of Unpaid Labor

（一）人机协作促进无酬劳动“价值升维”

The development of the platform economy shows that active users are both consumers of digital services and producers of data and content for digital platforms. From the perspective of human-machine relations, the supporting role of analytical AI is mainly reflected in increasing the exposure and attention of user-generated content through algorithmic recommendations, but its productive assistance to users in creating digital content is negligible. In other words, AI technology empowerment is actually “empowering the machine,” enabling ordinary users to achieve orderly presentations of “genuine expression” and “self-expression” in digital cyberspace, and in the process of stimulating likes, comments, reposts, tips, etc., helping digital capital harvest massive amounts of data and user attention. In contrast, a series of easy-to-use generative AI applications like ChatGPT greatly enrich the “toolbox” for user content generation. It is certain that “AI models will become ubiquitous, powerful assistants for every worker, placing a new type of hyper-personalized intelligence in people’s hands to increase productivity.” This is reflected in: under the human-machine collaboration model, on the one hand, instrumental tasks such as content production, optimization, and personalization in user-generated content can all be automatically completed by intelligent machines driven by generative AI technology; on the other hand, ordinary users can transcend the limitations of “technique” and “efficiency,” focusing exclusively on creative conception, content imagination, and self-expression. In this way, generative AI both “empowers the machine” and “empowers intelligence,” and low-threshold AI creation makes “artistic survival” a “new normal” for ordinary people. As a result, in terms of value creation, the content ecology of digital cyberspace improves, the digital content economy develops healthily, and the value contribution of users’ unpaid labor becomes increasingly prominent.

平台经济的发展表明，活跃用户既是数字服务消费者，也是数字平台数据和内容生产者。从人机关系来看，分析式人工智能的加持作用主要体现为通过算法推荐增加用户生成内容的曝光度和关注度，但对用户数字内容制作的生产助力则乏善可陈。换言之，AI技术赋能实际上是“赋机”，使普通用户能够在数字网络空间实现“真情流露”“自我表达”的有序呈现，并在激发点赞、评论、转发、打援等过程中，帮助数字资本收割海量数据和用户注意力。相比之下，ChatGPT等一系列易于使用的生成式人工智能应用程序极大地丰富了用户内容生成的“工具箱”。可以肯定的是，“人工智能模型将成为每位工作者无处不在的得力助手，将新型超个性化智能放在人们手中来提高生产力”。这体现在：在人机协作模式下，一方面，用户生成内容的内容制作、优化、个性化等工具性任务，都能交由生成式人工智能技术驱动的智能机器来自动完成；另一方面，普通用户得以跨越“技法”和“效能”的限制，专事于创意构思、内容想象和自我表达。如此一来，生成式人工智能既“赋机”又“赋智”，低门槛AI创作使“艺术化生存”成为普通人的“新常态”。其结果，在价值创造上，数字网络空间内容生态得以改善，数字内容经济得以良性发展，用户无酬劳动的价值贡献也就愈发彰显。

(2) Human-Machine Collaboration Gives Rise to “Market Differentiation” in Crowdsourced Labor

（二）人机协作催生众包劳动“市场分化”

Entering the stage of digital capitalism, a typical labor form in the global digital labor landscape is online crowdsourced labor, also known as “cloud labor.” Based on the level of professional skills required to complete digital tasks, crowdsourced labor is further subdivided into freelancers represented by Upwork and micro-labor represented by Amazon Mechanical Turk. In the online crowdsourcing economy, the “empowering the machine” effect of analytical AI is mainly reflected in breaking the spatiotemporal constraints of the labor market through algorithmic task allocation and algorithmic rewards and punishments, thereby either opening new channels for professionals to monetize “cognitive surplus” or providing new employment opportunities for marginalized and vulnerable groups.

进入数字资本主义阶段，全球数字劳动图谱中的一个典型劳动形态就是网络众包劳动，也称“云劳动”。依据完成数字任务所需要的专业技能程度，众包劳动又细分为以跨境客（Upwork）为代表的自由职业者和以亚马逊土耳其机器人（Amazon Mechanical Turk）为代表的微劳动。在网络众包经济中，分析式人工智能的“赋机”效应主要体现在通过算法派单和算法奖惩打破劳动力市场的时空束缚，进而要么给专业人士变现“认知盈余”开拓新渠道，要么给边缘弱势群体提供新的就业机会。

With the accelerated development of embedded and plug-in applications of generative artificial intelligence, the online crowdsourcing labor market will undergo profound changes, and the employment situation and value contribution of digital laborers will also experience a clear “market differentiation.” This is mainly because crowdsourcing labor is essentially cognitive labor, and its procedural, repetitive labor tasks will sooner or later be automated. Among them, crowdsourcing micro-labor is mostly low-threshold digital tasks, and digital capital will certainly seize the opportunity to hand them over to “the magic tool for achieving goals”—intelligent machines. Existing research shows that, compared with manual annotation, ChatGPT’s automatic annotation of tweets (including relevance, stance, topic, and frame detection) is not only more accurate but also costs less than $0.003 per annotation—about 20 times cheaper than Amazon Mechanical Turk. For freelancers with “unique skills,” although tasks with high repetition, regularity, and predictability, such as writing, translation, mapping, and code review, will shrink or even disappear, the labor returns for completing “one-of-a-kind” cognitive works will also become more generous. In practice, AI large models generate a rich capability matrix, helping digital creators break through capacity constraints, thereby having more time and energy to engage in more imaginative and challenging digital labor. According to Accenture’s forecast, by 2030, 75% of knowledge workers globally will interact daily with applications, services, or agents supported by foundation models. Furthermore, in human-machine dialogue, the massive knowledge content generated by intelligent machines under prompts and inducements is not without “insightful views” or “creative inspiration” that “open up the mind.” The result is that in the value production of digital capitalism, the upgrading of intelligent machines empowered by large models, on the one hand, leads to the loss of job opportunities for ordinary laborers engaged in “ghost work,” and on the other hand, increases wealth-creation opportunities for “digital nomads” skilled in creativity.

随着生成式人工智能嵌入式、插件式应用的加速发展，网络众包劳动力市场将迎来深刻变革，数字劳动者的就业境遇和价值贡献也将发生明显的“市场分化”。这主要是因为，众包劳动在本质上是一种认知劳动，其程序化、重复性劳动任务迟早会被自动化。其中，众包微劳动大都是低门槛的数字化任务，数字资本定将伺机交由“达到目的的神器”——智能机器去完成。已有的研究表明，相比人工标注，ChatGPT对推文（包括相关性、立场、主题和框架检测）的自动标注不仅更加准确，而且每条注释成本不到0.003美元——比亚马逊土耳其机器人便宜约20倍。对于拥有“独门秘技”的自由职业者而言，虽然撰文、翻译、制图、代码核查等重复性、规则性和预测性较高的任务也会萎缩甚至消失，但完成“独一无二”的认知作品的劳动回报也将更为丰厚。在实践中，AI大模型衍生出丰富的能力矩阵，助力数字创客们突破产能限制，从而有更多的时间和精力去从事更具想象力和挑战性的数字劳动。据埃森哲公司预测，到2030年，全球75%的知识工作者每天都要与由基础模型支持的应用程序、服务或代理进行互动。此外，在人机对话中，智能机器在提示和诱导下生成的海量知识内容不乏有令人“脑洞大开”的“真知灼见”抑或“创意灵感”。其结果就是，在数字资本主义价值生产中，大模型加持下的智能机器升级换代，这一方面导致从事“幽灵工作”的普通劳动者丧失了工作机会，另一方面为擅长创意的“数字游民”增加了创富机会。

(III) Human-Machine Collaboration Helps Improve the On-Demand Labor “Environment”

（三）人机协作助力改善按需劳动“环境”

Entering the digital economy era, the gig economy has developed rapidly, breaking through the constraints of time and space, giving rise to on-demand labor based on mobile applications (Apps). In essence, on the virtual production line “painstakingly” built by digital capital, gig workers are merely mobile data points controlled by the platform system, busy filling the gaps of “AI’s last mile.” From the perspective of human-machine relations, the technical empowerment of analytical artificial intelligence, on the one hand, helps ordinary laborers achieve low-threshold rapid employment, promoting digital inclusion and employment equity to a certain extent; on the other hand, platform-based enterprises vigorously implement “strictest algorithms,” deliberately stripping away “employer responsibility,” plunging many gig workers into physically and mentally exhausting employment instability. This results in a scenario once mentioned by Marx reappearing in the digital economy era, namely, “the machine possesses the wonderful power of reducing human labor and making it more effective, yet it causes hunger and excessive fatigue.” [10]10 Marx, *Capital*, Vol. 1, Chapter 15, Section 6, referring to the paradoxical effects of machinery under capitalism.

进入数字经济时代，零工经济突破时空束缚疾速发展，催生了基于移动应用程序（App）的按需劳动。从本质上讲，在数字资本“苦心孤诣”打造的虚拟生产流水线中，网约零工只是一个个受平台系统操控、忙于补位“AI最后一公里”的流动数据点而已。从人机关系来看，分析式人工智能的技术加持，一方面助力普通劳动者实现低门槛快速就业，在一定程度上促进了数字普惠和就业公平；另一方面，平台型企业极力推行“最严算法”，蓄意剥离“雇主责任”，使众多网约零工陷入身心俱疲的就业不稳定之中。其结果导致马克思曾提到过的场景在数字经济时代重现，即“机器具有减少人类劳动和使劳动更有成效的神奇力量，然而却引起了饥饿和过度的疲劳”。

In recent years, promoting algorithmic goodness and safeguarding the labor rights of gig workers has increasingly become a social consensus. Under pressure from government regulation and social pressure, digital capital also intends to refine algorithm management through platform system improvements, to quell the dissatisfaction of digital laborers and consolidate the platform production system. The platform embedding of generative artificial intelligence can push the gig economy from human-machine antagonism to human-machine collaboration, thereby improving the on-demand labor experience and easing labor-capital contradictions. This is because the intelligent upgrade of platform digital customer service, from “cold” formulaic responses to “friendly and relaxed” chat conversations, even allows laborers to customize “intimate companions” or “AI assistants.” This upgrade, on the one hand, can create a more pleasant and humane digital labor environment for laborers, and on the other hand, can change with the scenario, providing laborers with more personalized service prompts, process planning, and safety reminders in real time, thereby making the task completion process “effortless.” Furthermore, as native applications of generative artificial intelligence become increasingly abundant, laborers can also freely create “digital avatars” to maintain communication with multiple clients simultaneously, better catering to consumer preferences.

近年来，促进算法向善，维护网约零工的劳动权益日益成为社会共识。迫于政府监管和社会压力，数字资本也有意通过平台系统精进完善算法管理，以平息数字劳动者的不满，巩固平台生产体制。生成式人工智能的平台嵌入可以推动零工经济从人机对立转向人机协作，从而改善按需劳动体验，缓和劳资矛盾。这是因为，平台数字客服智能升级，从“冷冰冰”的程式化作答转变为“友好而轻松”的聊天对话，甚至允许劳动者量身定制“知心伙伴”“AI助理”。这种升级一方面可以给劳动者营造一个更加愉悦、合乎人性的数字劳动环境，另一方面也可以随场景而变，实时向劳动者提供更加个性化的服务提示、流程规划和安全提醒，从而使任务完成过程变得“轻而易举”。此外，随着生成式人工智能的原生应用日趋丰富，劳动者还可以随心打造“数字分身”，同时与多个客户保持沟通，以更好地迎合消费者偏好。

(IV) Human-Machine Collaboration Promotes “Job Transformation” in Hired Digital Labor

（四）人机协作推动雇佣数字劳动“工作转换”

Under the platform production system, digital capital pushes flexible employment practices to the limit: on the one hand, through the construction of multilateral market relations, it achieves “non-employment exploitation” of digital labor in the form of online crowdsourcing and outsourcing; on the other hand, digital capital owners spare no expense in recruiting global digital technology talent to engage in platform system infrastructure design, algorithm development, digital product R&D and operation, making them enterprise architects, algorithm engineers, data analysts, and operations engineers. These digital technology talents have multiple identities: they are both highly educated mental laborers and digital industrial workers, while also belonging to middle- and high-income groups. Broadly speaking, in the application scenarios of analytical artificial intelligence, the production contributions of these hired digital laborers are manifested in the data value chain of “analyzing data, discovering patterns, forming insights, and building predictions.”

在平台生产体制下，数字资本将弹性用工实践推向极限：一方面,通过多边市场关系构建，以网络众包、外包等形式实现对数字劳动的“非雇佣剥削”；另一方面，数字资本所有者又不惜重金招揽全球数字科技人才，从事平台系统基础架构设计、算法开发、数字产品研发与经营，使其成为企业架构师、算法工程师、数据分析师和运维工程师等。这些数字科技人才具有多重身份，他们既是高学历脑力劳动者，也是数字产业工人，同时又属于中高收入群体。大体上看，在分析式人工智能的应用场景中，这些雇佣数字劳动者的生产贡献彰显于“分析数据、发现模式、形成洞见、建立预测”的数据价值化链路之中。

With the accelerated arrival of the AI large model era, the automation of tasks in digital commodity production is accelerating. Driven by generative artificial intelligence technology, the work focus of tech geeks concentrated in leading platform-based enterprises will shift, and new job positions will emerge accordingly. Specifically, in terms of platform infrastructure construction, seeking new breakthroughs in new chipset architectures, hardware innovation, and efficient algorithms has become a key task for core employees of digital technology giants; in terms of AI large model R&D, model construction, pre-training, fine-tuning alignment, and inference deployment are closely linked, making algorithm engineers, prompt engineers, and other professions popular digital occupations; in terms of vertical applications, tasks such as training, fine-tuning, pruning, and distillation of industry-specific models give rise to new digital labor professions such as AI product managers, AI quality controllers, and AI ethics consultants. In short, focusing on the R&D and production of general-purpose intelligent machines is increasingly becoming the normal work of digital laborers.

随着AI大模型时代的加速到来，数字商品生产的任务自动化提级加速。在生成式人工智能的技术驱动下，聚集于头部平台型企业的技术极客们的工作重心将发生转移，新的工作岗位也会应运而生。具体而言，在平台基础设施建设方面，寻求新的芯片组架构、硬件创新和高效算法等方面的新突破，成为数字科技巨头核心员工着力完成的关键性任务；在AI大模型研发方面，模型构建、预训练、调优对齐和推理部署环环相扣，使算法工程师、提示工程师等职业成为热门的数字职业；在垂直应用方面，行业专有模型的训练、精调、剪枝、蒸馏等工作，催生人工智能产品经理、人工智能质控员、人工智能伦理顾问等新的数字劳动职业的诞生。总之，专注于通用智能机器的研发与生产日益成为数字劳动者的工作常态。

III. Large Model Competition and New Trends in Digital Capital Accumulation

三、大模型竞争与数字资本积累的新趋势

Marx pointed out: “A certain degree of capital accumulation appears as a condition for the specific capitalist mode of production, and the specific capitalist mode of production in turn causes accelerated capital accumulation.” [11]11 Marx, *Capital*, Vol. 1, Chapter 24, Section 1, on the relationship between capital accumulation and the capitalist mode of production. Entering the stage of digital capitalism, analytical artificial intelligence drove the platform revolution, giving rise to and solidifying “data enclosure,” algorithmic autocracy, and platform monopoly, achieving accelerated accumulation of digital capital in a very short time. Currently, with the upgrading, iteration, and accelerated implementation of generative artificial intelligence, capitalist production has entered the era of AI large models. AI large models are technology-, capital-, and talent-intensive industries. As shown in Table 2, large model competition solidifies the digital oligopoly pattern, and is causing the capital accumulation of digital capitalism to present a new trend of “intelligent computing monopoly.”

马克思指出：“一定程度的资本积累表现为特殊的资本主义的生产方式的条件，而特殊的资本主义的生产方式又反过来引起资本的加速积累。”进入数字资本主义阶段，分析式人工智能推动平台革命，催生和固化“数据圈地”、算法专制和平台垄断，在极短的时间内实现了数字资本的加速积累。当下，随着生成式人工智能的升级迭代和加速落地，资本主义生产进入了AI大模型时代。AI大模型是技术、资本和人才密集型产业。如表2所示，大模型竞争固化数字寡头格局，正在使数字资本主义资本积累呈现“智算垄断”的新趋势。

Image

图片

(I) Computing Power Competition Becomes a New Focus of Digital Capital Accumulation

（一）算力争夺成为数字资本积累新焦点

Entering the digital economy era, data, computing power, and algorithms form a “trinity,” becoming necessary resources for digital technology giants to seek and consolidate platform monopoly. With the application of AI content generation technology in digital commodity production, digital platforms, which use advanced algorithms as competitive weapons, are accelerating their transformation into high-performance computing platforms. In the new round of resource competition for digital capital, influenced by factors such as the diffusion of platform production systems, the surge in open databases, and the application of synthetic data, the importance of data resources has begun to decline. In contrast, creating foundation models is a complex, costly, and computationally intensive task; the computing power required to train top-tier AI models is growing exponentially.

进入数字经济时代，数据、算力和算法“三位一体”，成为了数字科技巨头谋取和巩固平台垄断的必要资源。随着人工智能内容生成技术在数字商品生产中的应用，以先进算法为竞争利器的数字平台正加速向高性能计算平台迈进。在数字资本的新一轮资源争夺中，受平台生产体制扩散、开放式数据库激增以及合成数据应用等因素的影响，数据资源的重要性开始下降。相比之下，创建基础模型是一项复杂、成本高昂的计算密集型工作；训练顶级人工智能模型所需的计算量正在呈现指数级增长。

Taking the AI language model GPT-3 as an example, this model has 175 billion parameters, uses a high-performance computing cluster composed of 10,000 V100 graphics processing units and 285,000 central processing units, takes 14.8 days for a single training run, and consumes a total computing power of approximately 3640 PF-days (that is, if one trillion calculations were performed per second, it would take 3640 days). This means that in the digital productive forces revolution driven by generative AI technology, the importance of intelligent computing power for training, fine-tuning, and deploying large models has become sharply prominent. It can be said that competition over AI large models is essentially competition over computing power; whoever possesses abundant advanced computing power will likely achieve a complete victory in the large-model competition.

以人工智能语言模型GPT-3为例，该模型参数量达1750亿，采用1万颗V100图形处理器和28.5万个中央处理器组成的高性能计算集群，单次训练用时14.8天，总算力消耗约为3640PF-days（即假如每秒计算一千万亿次，需要计算3640天）。这意味着，在生成式人工智能技术驱动的数字生产力革命中，用于训练、调优和部署大模型的智能算力的重要性急剧凸显。可以说，AI大模型竞争，实际上就是算力之争，谁拥有充沛的先进算力，谁就将大概率在大模型竞争中完胜。

(II) Digital oligopoly competition drives the monopolization of digitalized means of production to new heights

（二）数字寡头竞争推高数字化生产资料的垄断

The value movement of capitalism “presupposes the separation between the workers and the ownership of the conditions for the realization of their labour,” and “once capitalist production has taken root, it not only maintains this separation but reproduces it on an ever-expanding scale.” [1]1 Karl Marx, *Capital*, Vol. 1, Chapter 26. In terms of practical pathways, capital strives to drive the machine revolution, using “objectified knowledge power” to achieve the “deskilling” of labour and the expropriative appropriation of society’s “general intellect.” Entering the stage of digital capitalism, the establishment and consolidation of the platform production system has pushed the monopolization of digitalized means of production to new heights.

资本主义价值运动“以劳动者和劳动实现条件的所有权之间的分离为前提”，“资本主义生产一旦站稳脚跟，它就不仅保持这种分离，而且以不断扩大的规模再生产这种分离”。从实践路径来看，资本竭力推动机器革命，以“对象化的知识力量”来实现劳动“去技能化”和社会“一般智力”的剥夺性占有。进入数字资本主义阶段，平台生产体制的建立和巩固将数字化生产资料的垄断推向了新的高度。

At present, large models are competing fiercely, a scene of great bustle, but the engineering combination of “big data + large computing power + strong algorithms” means that the true players are few and far between. “Apart from the world’s top enterprises, almost no organization can complete this task on its own; it exceeds the capabilities and methods at their disposal.” [2]2 Kai-Fu Lee, *AI Superpowers: China, Silicon Valley, and the New World Order* (Houghton Mifflin Harcourt, 2018), p. 112. Therefore, although “AI startups” are everywhere, they can only become “barnacles on the hull” of the large tech companies.

当下，大模型百舸争流，热闹非凡，但“大数据+大算力+强算法”的工程化结合意味着真正的玩家屈指可数。“除了全球顶级企业外，几乎所有的组织都无法仅凭一己之力完成该任务，这超出了他们所掌握的能力和方法。”因此，尽管“AI初创公司”比比皆是，但它们也只能成为大型科技公司“船体上的藤壶”而已。

To seize business opportunities in the AI “arms race,” digital oligarchs are competing to restructure their platform ecosystems. On the one hand, digital oligarchs apply generative AI to every product, service, and business process they operate. In March 2023, Microsoft integrated generative AI technology into its product matrix, launching Microsoft 365 Copilot, which can not only directly generate Word, PPT, and Excel spreadsheets based on brief user instructions but also help users complete tasks such as summarizing meeting notes and processing emails. Google followed suit, integrating the upgraded Bard into its own office software “family bucket” and creating an office assistant called “Duet AI.” On the other hand, by building generative AI application super-apps, digital oligarchs embed AI large models as core infrastructure into numerous industries such as healthcare, medicine, education, logistics, credit, and entertainment. As a result, digital capital not only further strengthens its domination and control over the platform ecosystem but also, through oligopolistic monopoly over cognitive means of production, pushes the monopolization of the means of production that sustains the capitalist mode of production to its extreme. The outcome is that, from the perspective of economic development stages, digital capitalism is accelerating its evolution toward “AI capitalism,” and the dependence of the entire society’s production and life on platforms will grow day by day.

为了在AI“军备竞赛”中抢占商机，数字寡头们竞相重构平台生态系统。一方面，数字寡头将生成式人工智能应用到其自营的每个产品、服务和业务流程中。2023年3月，微软将生成式人工智能技术集成到产品矩阵中，推出了Microsoft 365 Copilot，它不仅可以根据用户简短指令直接生成Word、PPT和Excel电子表格，还可以帮助用户完成整理会议摘要、处理邮件等多项任务。谷歌紧随其后，将升级版Bard整合到自家办公软件“全家桶”中，并创造了名为“Duet AI”的办公助手。另一方面，数字寡头通过打造生成式人工智能应用超级App，将AI大模型作为核心基础设施嵌入健康、医疗、教育、物流、信贷、文娱等众多行业。由此，数字资本不仅进一步强化了对平台生态圈的支配和控制，而且通过对认知性生产工具的寡头独占，将维系资本主义生产方式的生产资料垄断推向极致。其结果就是，从经济发展阶段来看，数字资本主义正加速向“人工智能资本主义”演进，整个社会生产生活对平台的依赖也将与日俱增。

(III) Large models create a “surplus proletariat”

（三）大模型制造“过剩无产者”

Under the capitalist mode of production, “as soon as a means of labour appears as a machine, it immediately becomes a competitor of the worker himself.” [3]3 Karl Marx, *Capital*, Vol. 1, Chapter 15. During the period of industrial capitalism, the capitalist machine revolution went through stages of steam power, electrification, and informatization; the devaluation and replacement of labour by machines (systems) has always been a core issue of capital’s domination over labour. But overall, technological progress continuously gave rise to new job positions and new forms of labour, so that the development of the capitalist relative surplus population never spiraled “out of control.”

在资本主义生产方式下，“劳动资料一作为机器出现，立刻就成了工人本身的竞争者”。在工业资本主义时期，资本主义机器革命历经蒸汽机化、电汽化、信息化，机器（体系）对劳动的贬值和替代始终是资本宰制劳动的核心议题。但总体上看，技术进步不断催生新的工作岗位和新的劳动形态，从而使资本主义相对过剩人口发展始终没有走向“失控”。

After the advent of AI technology, both “labour enhancement” and “labour substitution” effects have coexisted. At present, with the accelerated commercial application and industrial development of generative AI technology, the technological unemployment caused by “machines replacing humans” seems to be taking on an overwhelming momentum. In the short term, although, as mentioned earlier, the technological empowerment and human-machine collaboration of generative AI have generally promoted the high-level evolution of digital labour, it should also be recognized that a large number of ordinary workers, unable to adapt to the accelerated changes in digital technology, are being reduced to a “surplus proletariat.” In the long run, even setting aside considerations of “superintelligence,” “social exclusion,” and “useless class,” under the onslaught of rapidly advancing generative AI technology, tasks or work unique to humans, such as knowledge creation, will be increasingly encroached upon by ever more powerful intelligent machines. As some scholars have noted, generative AI, “while becoming a tool for labour, may also become a ‘monster’ that ‘devours’ people’s digital labour.” [4]4 See, for example, Nick Srnicek and Alex Williams, *Inventing the Future: Postcapitalism and a World Without Work* (Verso, 2015), p. 171. The result is that intelligent production intensifies the problem of “technological unemployment,” placing ordinary workers such as the “digital poor” in a more challenging and hopeless situation, and the production order of digital capitalism will face a major shock.

人工智能技术问世后，“劳动增强”和“劳动替代”效应兼而有之。当下，生成式人工智能技术加速商业化应用和产业化发展，“机器换人”引发的技术性失业似乎呈现碾压之势。短期来看，虽然如前文所述，生成式人工智能的技术赋能和人机协作总体上推动了数字劳动的高阶演化，但同时也应该看到，大量普通劳动者因无法适应数字技术的加速变革而沦落成为“过剩无产者”。从长远来看，即使撇开“超级智能”“社会排斥”“无用阶级”等不论，在生成式人工智能技术突飞猛进的攻势下，人类专属的知识创造等工作或任务将不断被日益强大的智能机器所蚕食。正如有的学者所言，生成式人工智能“在成为劳动利器的同时，也可能成为‘吞噬’人的数字劳动的‘怪兽’”。其结果就是，智能化生产推动“技术性失业”问题日趋尖锐，将“数字穷人”之类普通劳动者置于更具挑战性、更看不见希望的境地，数字资本主义生产秩序将遭遇重大冲击。

IV. AI Technology Acceleration and New Variables in the Capitalist Production Order

四、AI技术加速与资本主义生产秩序的新变数

“Constant revolutionising of production, uninterrupted disturbance of all social conditions, everlasting uncertainty and agitation distinguish the bourgeois epoch from all earlier ones.” [5]5 Karl Marx and Friedrich Engels, *The Communist Manifesto*, Chapter 1. From the perspective of the capitalist application of science and technology, “the impact of technological acceleration on social reality is undoubtedly enormous.” [6]6 Hartmut Rosa, *Social Acceleration: A New Theory of Modernity* (Columbia University Press, 2013), p. 71. Entering the era of the digital economy, digital technological progress, driven by capital, exhibits a “flywheel effect,” making the capitalist production order more fraught with variables, turbulence, and even crisis. At present, the emergence and iteration of generative AI make the acceleration of capitalist technology even more unstoppable; digital capital pursues “general artificial intelligence,” but at the same time, it brings new variables to the capitalist production order.

“生产的不断变革，一切社会状况不停的动荡，永远的不安定和变动，这就是资产阶级时代不同于过去一切时代的地方。”从科学技术的资本主义应用来看，“科技加速对社会现实的影响无疑是巨大的。”进入数字经济时代，资本裹挟下的数字科技进步呈现出“飞轮效应”，使资本主义生产秩序更加充满变数、动荡，甚至陷入危机。当下，生成式人工智能的出现与迭代使资本主义技术加速更加势不可当，数字资本追逐“通用人工智能”，但同时也给资本主义生产秩序带来新的变数。

(I) AI technology acceleration hampers digital capital’s “spatio-temporal fix”

（一）AI技术加速掣肘数字资本“时空修复”

Marx pointed out that capital, on the one hand, “strives to reduce labour time to a minimum, while on the other hand it posits labour time as the sole measure and source of wealth.” [7]7 Karl Marx, *Grundrisse*, Notebook VII, “The Fragment on Machines.” Since the First Industrial Revolution, capital has successfully absorbed a great deal of “living labour” through repeated “spatio-temporal fixes” to offset the impact on capital’s rule of the antagonistic development between the “expansion of the wealth system” and the “shrinkage of the value system.” Entering the era of the digital economy, digital capital has constructed a platform production system, greatly expanding capital’s capacity for “spatio-temporal fixes.” In summary, digital capital disciplines digital labour to achieve the absorption of “living labour” on a larger scale; at the same time, digital capital promotes “digital survival” to more thoroughly realize the expropriation of society’s “general intellect.”

马克思指出，资本一方面“竭力把劳动时间缩减到最低限度，另一方面又使劳动时间成为财富的唯一尺度和源泉”。自第一次工业革命以来，资本通过一次次的“时空修复”成功地吸纳了很多“活劳动”来对冲“财富系统膨胀”和“价值系统萎缩”的对立发展给资本统治造成的冲击。进入数字经济时代，数字资本构筑平台生产体制，极大地拓展了资本的“时空修复”能力。归结起来，数字资本规训数字劳动，以在更大范围内实现对“活劳动”的吸纳；同时，数字资本推动“数字化生存”，以更加彻底地实现对社会“一般智力”的攫取。

At present, the accelerated application of generative AI technology releases enormous digital productive forces, creating an abundant outpouring of digital wealth with astonishing efficiency; but at the same time, it achieves a more thorough replacement of “living labour” by “dead labour” at an unimaginable speed. Thus, paradoxically, the more developed the productive forces of digital capitalism, the more inescapable the nightmare of the “singularity approach” in capital’s value movement becomes. Moreover, under the coercive law of competition, as the application of AI large models becomes low-code and democratized, the organic composition of capital in the global knowledge industry accelerates toward convergence, and digital capital’s capacity for the “spatio-temporal fix”—building a “digital empire” and exploiting and transferring surplus value—is continuously eroded. The result is that digital capital, mired in the AI-driven digital vortex, is rapidly losing its blood-sucking ability to “suck living labour”—the very foundation of its existence—amid its obsession with digital technology.

目前来看，生成式人工智能技术的加速应用释放出巨大的数字生产力，以令人惊叹的效率创造了数字财富的充分涌流；但同时，它又以超乎想象的速度实现“死劳动”对活劳动的更为彻底的替代。如此一来，吊诡的是，数字资本主义生产力越是发达，资本价值运动中“奇点逼近”的梦靥就越是挥之不去。并且，在竞争的强制规律作用下，随着AI大模型应用的低代码化、平民化，全球知识产业的资本有机构成加速趋同，数字资本构建“数字帝国”、盘剥和转移剩余价值的“时空修复”能力也被不断地消解。其结果就是，数字资本深陷AI化数字漩涡，“安身立命”的嗜血能力——“吮吸活劳动”——反而在对数字技术的迷恋中加速丧失。

(II) AI technology acceleration impacts the realization of the value of knowledge commodities

（二）AI技术加速冲击知识商品价值实现

As is well known, under market economy conditions, knowledge commodities such as software, databases, and product designs exhibit a significant feature of zero marginal cost, which makes their value realization subject to greater uncertainty. The establishment of the capitalist intellectual property system effectively repels various "anti-value" forces, thereby playing a key role in maintaining the capitalist production order. In the era of the digital economy, the self-valorization of digital capital presupposes the deliberate dissolution of individual intellectual property rights in user-generated content. The real-time, personalized, and contextualized nature of digital commodity production allows digital capital to reap substantial profits even without the shelter of the copyright system. However, overall, even with the support of analytical artificial intelligence technology, digital capital has not broken through the existing interest structure to apply "de-copyrightization" to professionally generated content in the same way, and thus has not yet fundamentally impacted the value realization of capitalist knowledge commodities.

众所周知，在市场经济条件下，软件、数据库、产品设计等知识商品具有显著的零边际成本特征，这种特征使其价值实现面临更多变数。资本主义知识产权制度的创设有效地拒止了各种“反价值”力量，进而在维系资本主义生产秩序方面发挥着关键性作用。进入数字经济时代，数字资本的自我增殖以蓄意消解用户生成内容的个人知识产权权益为前提。数字商品生产的实时化、个性化和场景化使数字资本甚至无需版权制度庇荫，就能赚得盆满钵满。但总体上看，即使有分析式人工智能技术的加持，数字资本也并没有冲破既有的利益格局来对专业生成内容如法炮制“去版权化”，因而尚未对资本主义知识商品价值实现造成根本冲击。

In contrast, the widespread adoption of generative artificial intelligence is rapidly and overwhelmingly impacting the existing capitalist copyright legal system. This is because AI large models, relying on their increasingly powerful learning capabilities, have nearly swept up and consumed all knowledge achievements in human history, producing an unlimited quantity of "not-real-but-seemingly-real" AI-generated content with ultra-high efficiency. For example, a human illustrator needs at least 2–3 days to complete a single work, while an intelligent drawing tool like Midjourney can create four works in just 10 seconds. Faced with the "ocean of machine-generated text" produced by AI large models that is both familiar and ambiguous, the current capitalist intellectual property system finds itself inadequate, even at a loss, in terms of tracing rights and identifying infringements. Generative AI can also create AI-cloned characters through style imitation, voice synthesis, and posture synthesis, rapidly accumulating users and harvesting traffic with an overwhelming and creative supply of digital content. As a result, under the impact of AI’s massive imitation and traffic diversion, the intellectual property value of existing rights holders and real-world content producers is greatly diluted.

相比之下，生成式人工智能的大行其道正在以迅雷不及掩耳之势冲击资本主义现存版权法体系。这是因为，AI大模型凭借自身日益强大的学习能力，几乎网尽和吞噬人类有史以来的一切知识成果，超高效地无限量生产出“非真似真”的人工智能生成内容。譬如，人类插画师至少需要2—3天的时间才能完成一幅作品，而中道（Midjourney）这样的智能绘图工具只需要10秒钟就可以创作4幅作品。面对AI大模型制造似曾相识但又似是而非的“机器文本海洋”，资本主义现行知识产权制度在溯源确权、侵权指认上显得力不从心，甚至无所适从。生成式人工智能还能通过风格模仿、语音合成和姿态合成打造AI版克隆人物，以令人应接不暇、创意十足的数字内容供给在极短的时间内积聚用户、收割流量。由此，在AI海量摹写和引流冲击下，既有作品权利人、现实内容生产者的知识产权价值被极大地稀释。

(3) AI Technology Accelerates and Intensifies Power Conflicts Between Government and Enterprises

（三）AI技术加速加剧政企权力冲突

Power is based on control over information and its dissemination. In the era of the digital economy, digital capital accumulates platform power through increasingly sophisticated algorithmic systems, building "digital empires." Practice shows that with the formation and consolidation of platform monopolies, the exercise of digital capital’s power is no longer confined to the sphere of market exchange but permeates every aspect of human production and life through digital surveillance and information manipulation, quietly playing the role of a "private government" and becoming a veritable "Leviathan" in the platform society. [12]12 "Leviathan" refers to Thomas Hobbes’s concept of the sovereign state as an all-powerful entity; here it is used critically to describe digital capital’s usurpation of state-like power. "Platform companies, as power subjects, have gradually come to the forefront, dissolving and reconstructing the state-led power structure, accelerating the flow of power from the state to the market." Therefore, once the bourgeois government becomes aware of the "power usurpation" inadvertently achieved by digital oligarchs, it becomes highly vigilant and prepares to retaliate. Fundamentally, in the game between government and enterprises in digital capitalism, digital capital’s acquisition of power originates from the infrastructuralization of digital media. Since the platform revolution, with the support of analytical AI applications, the generation of digital capital’s power has begun by linking supply and demand in multi-sided markets and controlling the precise distribution of digital content. Subsequently, the influence of digital capital’s power is manifested through platform "empowerment" that shapes the online behavior of market entities, and through personalized recommendations that affect user experience and psychological cognition. Moreover, the active users subjugated to the "digital empire" number in the hundreds of millions, willingly becoming "digital tenants," which means that the "governmentalization" of digital capital’s private power cannot be underestimated.

权力是基于对信息及其传播的控制。进入数字经济时代，数字资本凭借日益精进的算法系统积聚平台权力，打造“数字帝国”。实践表明，随着平台垄断的形成和巩固，数字资本的权力施展不再局限于市场交换领域，而是通过数字监视和信息操控渗透至人类生产生活的方方面面，悄然扮演着“私政府”的角色，俨然成为平台社会里的“利维坦”。“平台公司作为权力主体逐渐走到前台，对国家主导的权力结构产生消解和重构，加速了权力从国家流散到市场。”因此，面对数字寡头不经意间实现的“权力僭越”，资产阶级政府一旦有所醒悟，就会倍加警惕并酝酿回击。从根本上讲，在数字资本主义政企博弈中，数字资本的权力获取源自数字媒介基础设施化。自平台革命出现以来，在分析式人工智能的应用支持下，数字资本的权力生成发轫于链接多边市场供需，操控数字内容精准分发。进而，数字资本的权力影响体现为通过平台“赋能”左右市场主体的在线行为，通过个性化推送，影响用户体验和心理认知。并且，臣服于“数字帝国”的活跃用户动辄数以亿计，自愿甘当“数字佃农”，这意味着数字资本私权“政府化”不容小觑。

Practice shows that for the capitalist production order to operate "smoothly," it requires the government to use public power to construct a "social embedding" mechanism to alleviate labor-capital contradictions. Currently, with the accelerated deployment of generative AI, digital capital has greatly consolidated the power foundation of the "digital empire" by redefining user experience and reshaping the platform service ecosystem. More critically, digital capital can not only firmly control the diffusion of knowledge-based content through intelligent dissemination but also directly acquire the "new power" of value shaping and concept production. That is, digital capital "sets its own formats, standards, and value orientations for knowledge production through model design and corpus selection, becoming the controller of knowledge production." Thus, digital capital skillfully infiltrates the "private goods" of profit-seeking concepts into the natural and fluent output of AI-generated content, while the public, enamored with the AI experience, is manipulated and persuaded without being aware of it. As some scholars have pointed out: "Once the two major capabilities of AI in content production and content distribution form a closed loop, its dominance over information flows and control over consumers will be immensely powerful." In this way, digital capital can provide information and knowledge through ultra-personalized customization, wielding the power of interpretation and discourse to reshape the capitalist knowledge order, thereby encroaching on the ideological power of the bourgeois state, hollowing out the government’s "toolbox" for social soft governance, and threatening the capitalist political order. For this reason, it is only to be expected that the struggle for influence between government and enterprises in digital capitalism is intensifying.

实践表明，资本主义生产秩序要保持“顺畅”运行，离不开政府运用公权力构建“社会嵌入”机制，缓和劳资矛盾。当下，随着生成式人工智能的加速落地，数字资本通过重新定义用户体验和重塑平台服务生态，极大地巩固了“数字帝国”的权力基础。更为关键的是，数字资本不仅可以通过智能传播来牢牢掌控知识性内容的扩散，还能直接获取价值观塑造和观念生产的“新权力”。也即，数字资本“通过模型的设计和语料的选择，为知识生产设定出自己的格式、标准和价值取向，成为知识生产的控制者”。于是，数字资本通过自然流畅的人工智能生成内容输出，巧妙地将逐利观念的“私货”渗揉其中，而迷恋于AI体验的社会大众被操纵、说服却难以自知。正如有学者所指出的：“AI在内容生产和内容分发上的两大能力一旦形成闭环，对信息流的主导以及对消费者的掌控将威力巨大。”如此一来，数字资本就可以凭借超个性化的定制方式提供信息和知识，以坐拥解释权和话语权，重塑资本主义知识秩序，进而蚕食资产阶级国家意识形态权力，掏空政府社会软治理“工具箱”，威胁资本主义政治秩序。正因如此，数字资本主义政企之间的影响力争夺加剧，也就在意料之中了。

V. Conclusion

五、结语

With the iterative upgrading and daily application of generative AI such as GPT-4, Gemini, and Claude, human production and life have entered the era of AI large models, and the production process and production relations of digital capitalism are undergoing revolutionary and subversive changes. Overall, the technical deployment and production operation of generative AI are initiating a new industrial revolution, and the mental production of digital capitalism has entered a new stage of "ultra-personalized customization and one-click generation." At the same time, the advent of the AI large model era has accelerated the democratization of AI technology, and the human-machine relationship in digital capitalism is undergoing complex and frequent changes. However, the increasing "human-like intelligence" and autonomy of intelligent machines have not ended digital labor; on the contrary, human-machine collaboration frees people from repetitive and tedious labor tasks, allowing them to focus on more creative work. Furthermore, under "Silicon Valley private ownership," generative AI, on the one hand, efficiently creates a boom in digital wealth, while on the other hand, it exacerbates platform monopolies, relative surplus population, and power conflicts between government and enterprises, thereby making the inherent contradictions of the value movement of digital capitalism increasingly difficult to reconcile. Only by transcending the logic of capital, fundamentally transforming the platform system of digital capitalism, and exploring the construction of digital socialism can we truly achieve AI universal benefit and the liberation of digital labor.

随着GPT-4、Gemini、Claud等生成式人工智能的升级迭代和日常化应用，人类生产生活进入了AI大模型时代，数字资本主义生产过程和生产关系正在发生革命性、颠覆性的变化。总体上看，生成式人工智能的技术落地和生产运维，正在开启一场新的工业革命，数字资本主义的精神生产迈进“超个性化定制、一键式生成”新阶段。与此同时，AI大模型时代的到来加速了AI技术的平民化，数字资本主义人机关系正在发生复杂频变。然而，智能机器的“类人智能”和自主性的日益增强并没有结终数字劳动，相反，人机协作将人们从重复且繁琐的劳动任务中解脱出来，专注于更具创造性的工作内容。进一步地，在“硅谷私有制”下，生成式人工智能一方面高效制造了数字财富繁荣，另一面加剧了平台垄断、相对人口过剩和政企权力冲突，进而使数字资本主义价值运动的固有矛盾愈发难以调和。只有超越资本逻辑，彻底变革数字资本主义平台体制，探索建设数字社会主义，才能真正实现AI普惠和数字劳动解放。

Online Editor: Tongxin

网络编辑：同心