当算法成为新主人：当代奴役理论的演进与批判
When Algorithms Become the New Master: The Evolution and Critique of Contemporary Slavery Theory
摘要 / Abstract
本文系统梳理了2025—2026年人文社会科学领域关于“奴役”的最新理论进展。传统奴役概念建立在人身占有与暴力强制之上，而当代奴役理论的核心发现是：奴役的本质不在暴力，而在“意志的屈从”。从新罗马主义对“工资奴役”的重新论证，到数字平台中“数据异化”与“数字奴役”的提出，再到技术封建主义对“云农奴”处境的刻画，理论界正在构建一条从工业资本主义到数字资本主义的奴役形态演进链。本文认为，当代奴役理论最关键的贡献在于揭示了一种“隐形化机制”——技术外衣、合法性外观与算法控制共同将新型奴役关系从法律范畴中抹去，使之成为一种“看不见的锁链”。
This paper systematically reviews the latest theoretical developments in the humanities and social sciences concerning “slavery” from 2025 to 2026. Traditional conceptions of slavery were built upon physical ownership and violent coercion, whereas the core finding of contemporary slavery theory is that the essence of slavery lies not in violence but in “subjection of the will.” From the neo-Roman re-argumentation of “wage slavery,” to the concepts of “data-based alienation” and “digital slavery” in digital platforms, and further to the depiction of the “cloud serf” condition under techno-feudalism, theorists are constructing an evolutionary chain of slavery forms from industrial capitalism to digital capitalism. This paper argues that the most critical contribution of contemporary slavery theory lies in revealing an “invisibilisation mechanism”—technological veneer, appearance of legality, and algorithmic control collectively erase new forms of slavery from legal categories, rendering them an “invisible chain.”
关键词：奴役理论；数字劳动；算法控制；技术封建主义；异化
Keywords: slavery theory; digital labour; algorithmic control; techno-feudalism; alienation
一、引言：重新审视“奴役” / 1. Introduction: Re-examining “Slavery”
当我们谈论“奴役”时，脑海中浮现的往往是锁链、鞭子和种植园。这种印象并非错误，但它遮蔽了一个更深层的问题：奴役的本质究竟是什么？
When we talk about “slavery,” what often comes to mind are chains, whips, and plantations. This impression is not wrong, but it obscures a deeper question: what is the essence of slavery?
2026年，牛津大学出版社出版的《薪酬不足与不被承认：国家引发的非自由劳动》一书对这一常识发起了系统挑战。该书作者Katie Bales与Amir Paz-Fuchs指出，现有法律对“非自由劳动”的理解主要被“现代奴役”话语所框定，这种狭窄的路径遮蔽了产生劳动不自由的更广泛结构性条件，限制了法律识别和解决剥削问题的能力。他们主张，法律范畴中的“奴隶制”“奴役”和“强迫劳动”是在殖民议程中被塑造出来的，这意味着自由主义法律秩序至今仍在忽视大量当代形式的强制劳动关系。
In 2026, the book Underpaid and Unrecognized: State-Induced Unfree Labour, published by Oxford University Press, launched a systematic challenge to this common sense. Its authors, Katie Bales and Amir Paz-Fuchs, point out that existing legal understandings of “unfree labour” are mainly framed by the discourse of “modern slavery.” This narrow path obscures the broader structural conditions that produce unfree labour and limits law’s capacity to identify and address exploitation. They argue that the legal categories of “slavery,” “servitude,” and “forced labour” were shaped within a colonial agenda, meaning that the liberal legal order continues to overlook a large number of contemporary forms of coercive labour relations.
这一批判指向一个关键问题：如果法律定义本身就有历史局限性，那么，当代的“奴役”是否正在以一种我们尚未学会识别的方式发生？
This critique points to a key question: if legal definitions themselves have historical limitations, then is contemporary “slavery” occurring in ways we have not yet learned to recognise?
二、工资奴役：重新激活一个被遗忘的概念 / 2. Wage Slavery: Reactivating a Forgotten Concept
要回答上述问题，理论界首先做的是回到起点——工资劳动本身是否构成一种奴役？
To answer the above question, theorists first return to the starting point: does wage labour itself constitute a form of slavery?
2026年发表于政治理论期刊的论文《工资奴役：一种新罗马主义的论证》为这个充满争议的概念提供了迄今为止最严密的哲学辩护。作者从新罗马政治思想传统中打捞出一种对奴隶制的理解：奴隶制的本质不是财产所有或社会性死亡，而是“屈从于他人的意志”。激进共和主义者曾经用这一框架批判雇佣劳动关系，呼吁财产社会化。
A 2026 article in a political theory journal, “Wage Slavery: A Neo-Roman Account,” offers the most rigorous philosophical defence of this controversial concept to date. The author retrieves from the neo-Roman tradition an understanding of slavery: the essence of slavery is not property ownership or social death, but “subjection to another’s will.” Radical republicans once used this framework to criticise the employment relationship and called for the socialisation of property.
但该论文也设置了一个重要限定：工资奴役的指控不应被泛化。作者论证，只有当“劳动者满足其最基本物质需求的能力依赖于雇主的意志”时，工资奴役才成立。换言之，工资奴役不是所有雇佣关系的代名词，而是一种极端的经济不自由状态。
However, the article also sets an important qualification: the charge of wage slavery should not be generalised. The author argues that wage slavery exists only when “a worker’s ability to meet their most basic material needs depends on the will of the employer.” In other words, wage slavery is not a synonym for all employment relations, but an extreme condition of economic unfreedom.
这一区分至关重要。它回应了长期以来对“工资奴役”概念的合理怀疑——正如哲学家Tommy Shelby所言：“我有雇主，缺乏足够财富以便不替人工作就能维持生计。但没有任何有意义的感觉说我是奴隶，说我是奴隶会侮辱我那些被奴役的祖先”。新罗马主义框架的贡献在于，它不否认工资劳动者与动产奴隶之间的差异，而是通过重新定义“奴役”的内涵——从“被占有”转向“被支配”——来揭示两者之间被忽视的结构性连续性。
This distinction is crucial. It responds to long-standing reasonable doubts about the concept of “wage slavery”—as the philosopher Tommy Shelby once put it: “I have an employer, and I lack sufficient wealth to sustain my life without working for someone else. But there is no meaningful sense in which I am a slave, and calling me a slave would insult my enslaved ancestors.” The contribution of the neo-Roman framework is that it does not deny the differences between wage labourers and chattel slaves; rather, by redefining “slavery” from “being owned” to “being dominated,” it reveals the overlooked structural continuity between the two.
三、数字奴役：当异化获得新的形态 / 3. Digital Slavery: When Alienation Takes a New Form
如果说工资奴役理论重新审视的是工业资本主义的基本关系，那么“数字奴役”理论则将矛头指向了平台经济与人工智能产业。
If wage slavery theory re-examines the basic relations of industrial capitalism, then “digital slavery” theory directs its critique at the platform economy and the artificial intelligence industry.
2026年发表于《纳米伦理》期刊的论文《AI凝视的背后：生成式人工智能、微工作者与数字奴役》提出了一个极具冲击力的判断：生成式AI的崛起并非因为机器获得了接近人类的智能，而是因为一种“不可思议的计算能力”让AI“看起来智能”。支撑这种“看起来智能”的，是大量看不见的人类劳动——数据标注、内容筛选、算法微调。这些工作由所谓的“微工作者”完成，他们往往来自发展中国家，在恶劣条件下工作，在工资、病假、假期、工会保护和隐私方面的权利受到严格限制。
A 2026 article in the journal NanoEthics, “Behind the AI’s Gaze: Generative Artificial Intelligence, Microworkers and Digital Slavery,” makes a striking judgment: the rise of generative AI is not because machines have acquired intelligence close to that of humans, but because an “incredible computational capacity” makes AI “appear intelligent.” What supports this “appearance of intelligence” is a large amount of invisible human labour—data annotation, content filtering, and algorithm fine-tuning. These tasks are performed by so-called “microworkers,” often from developing countries, working under harsh conditions and with strictly limited rights regarding wages, sick leave, holidays, union protection, and privacy.
该论文的核心理论贡献在于提出了“基于数据的异化”这一概念。马克思所描述的工人与其劳动产品的异化，在数字时代获得了新的双重维度：工人与其数据的异化——数据被数字平台用于组织劳动；以及工人与自身的异化——平台对数据的控制削弱了工人的自主性。
The article’s core theoretical contribution lies in proposing the concept of “data-based alienation.” The alienation of the worker from the product of their labour, as described by Marx, acquires a new dual dimension in the digital age: the worker’s alienation from their data—data are used by digital platforms to organise labour; and the worker’s alienation from themselves—platform control over data weakens workers’ autonomy.
更具颠覆性的是该论文对人与机器关系的分析。在正常的服务关系中，人类应被机器服务；但在数字平台的模型中，这一关系被颠倒了——人类服务于机器，且是为社会中有限一部分人的利益服务。论文进一步指出，在数字平台模式下，旧的奴役形式之所以不再被采用，仅仅是因为存在“更高效的个体剥削形式”，这些形式甚至可以在工人的“主动配合”下实现，并具有“合法性的外观”。
Even more subversive is the article’s analysis of the human–machine relationship. In a normal service relationship, humans should be served by machines; but in the digital platform model, this relationship is reversed—humans serve machines, and they do so for the benefit of a limited part of society. The article further points out that under the digital platform model, older forms of slavery are no longer adopted simply because “more efficient forms of individual exploitation” exist, forms that can even be realised with workers’ “active cooperation” and that possess an “appearance of legality.”
这指向了一个关键洞见：当代奴役的“进步”不在于奴役的消失，而在于它学会了穿上合法的外衣。
This points to a key insight: the “progress” of contemporary slavery does not lie in the disappearance of slavery, but in the fact that it has learned to put on the cloak of legality.
四、算法封建主义：奴役的“前现代”回归 / 4. Algorithmic Feudalism: The “Pre-Modern” Return of Slavery
如果数字奴役描述的是劳动过程中的控制，那么“技术封建主义”理论则试图从宏观政治经济学层面把握数字平台时代的权力结构。
If digital slavery describes control within the labour process, then “techno-feudalism” theory attempts to grasp the power structure of the digital platform era at the level of macro political economy.
这一理论以希腊前财政部长雅尼斯·瓦鲁法基斯和法国经济学家塞德里克·迪朗为代表，其核心主张是：当代资本主义正在向一种具有中世纪封建等级制特征的新型生产关系倒退。科技寡头以“数字领主”的姿态控制关键生产资料——数据、算法、算力，用户则沦为“数字农奴”，被迫让渡行为数据并接受算法规训。
This theory is represented by former Greek Finance Minister Yanis Varoufakis and French economist Cédric Durand. Its core claim is that contemporary capitalism is regressing towards a new type of production relation characterised by medieval feudal hierarchy. Tech oligopolies control key means of production—data, algorithms, and computing power—as “digital lords,” while users are reduced to “digital serfs,” forced to surrender behavioural data and submit to algorithmic discipline.
在这一框架中，一个关键机制是“平台佃租”。正如中世纪领主通过控制土地来向农奴收取地租，数字平台通过控制流量和数据的分配权来向用户和劳动者收取“数字地租”。有学者指出，数字领土分配通过算法黑箱实现流量资源的等级化垄断，重现了“领主控制土地”的权力结构。早期现代工作坊的历史类比也值得注意——这些机构在向工业资本主义过渡中发挥了重要作用，为农业社会崩溃产生的“过剩人口”准备工厂劳动。
In this framework, a key mechanism is “platform rent.” Just as medieval lords collected rent from serfs by controlling land, digital platforms collect “digital rent” from users and workers by controlling the distribution of traffic and data. Some scholars point out that digital territorial allocation achieves a hierarchical monopoly of traffic resources through algorithmic black boxes, reproducing the power structure of “lords controlling land.” The historical analogy of early modern workhouses is also noteworthy—these institutions played an important role in the transition to industrial capitalism, preparing the “surplus population” produced by the collapse of agrarian society for factory labour.
2026年发表于《AI与社会》的评论文章《反思AI的“解放”幻象》以更直白的语言表达了这一批判：“在企业的统治下，AI没有解开任何人的锁链，它把人绑得更紧了。你的劳动没有被解放，它被监视、抽取和审视到消失。用熟悉的硅谷术语来说，锁链没有消失，它们只是升级到了2.0版本”。
A 2026 commentary in AI & Society, “Rethinking AI’s ‘Liberation’ Fantasy,” expresses this critique in more direct language: “Under corporate rule, AI has not unlocked anyone’s chains; it has bound people tighter. Your labour has not been liberated; it has been surveilled, extracted, and scrutinised into disappearance. In familiar Silicon Valley terms, the chains have not disappeared; they have simply been upgraded to version 2.0.”
五、隐形化的机制：奴役如何“消失”在法律视野中 / 5. The Mechanism of Invisibilisation: How Slavery “Disappears” from Legal View
上述三支理论力量——新罗马主义的工资奴役论、数字劳动批判、技术封建主义——共同指向一个核心问题：当代奴役为何如此难以被识别和命名？
The above three theoretical forces—neo-Roman wage slavery theory, digital labour critique, and techno-feudalism—jointly point to a core question: why is contemporary slavery so difficult to identify and name?
第一重隐形化来自法律范畴的历史包袱。如前所述，Bales和Paz-Fuchs揭示，自由主义法律秩序中的“奴隶制”“奴役”和“强迫劳动”定义是在殖民议程中形成的，这使得法律体系对当代强制劳动关系的识别能力存在结构性缺陷。他们因此提出了一个“光谱式方法”，用三维框架来识别剥削性劳动实践的关键维度，并主张马克思主义女性主义分析框架比自由主义法律模型更能捕捉劳动的“不自由”。
The first layer of invisibilisation comes from the historical baggage of legal categories. As mentioned earlier, Bales and Paz-Fuchs reveal that the definitions of “slavery,” “servitude,” and “forced labour” in the liberal legal order were formed within a colonial agenda, which creates a structural defect in the legal system’s capacity to identify contemporary coercive labour relations. They therefore propose a “spectrum approach,” using a three-dimensional framework to identify the key dimensions of exploitative labour practices, and argue that a Marxist-feminist analytical framework can better capture the “unfreedom” of labour than liberal legal models.
第二重隐形化来自技术外衣。数字平台通过“算法管理”取代了传统的监工，通过“动态定价”取代了固定的工资谈判，通过“用户评分”取代了直接的纪律惩罚。这些机制表面上看起来只是“技术中立”的效率工具，实际上构成了一个“数字全景监狱”——一种通过数据化和监控恐惧而非直接强制来规训工人的架构。
The second layer of invisibilisation comes from the technological veneer. Digital platforms replace traditional overseers with “algorithmic management,” replace fixed wage bargaining with “dynamic pricing,” and replace direct disciplinary punishment with “user ratings.” On the surface, these mechanisms appear to be merely “technologically neutral” efficiency tools; in reality, they constitute a “digital panopticon”—an architecture that disciplines workers through datafication and the fear of surveillance rather than direct coercion.
第三重隐形化来自合法性外观。数字奴役理论指出，当代剥削之所以不再采用传统奴役形式，是因为存在“更高效的个体剥削形式”，这些形式“甚至可以在工人的主动配合下实现，并具有合法性的外观”。工人签署了合同、接受了条款、点击了“同意”——形式上的自愿恰恰构成了实质上的不自由。
The third layer of invisibilisation comes from the appearance of legality. Digital slavery theory points out that contemporary exploitation no longer adopts traditional slavery forms because “more efficient forms of individual exploitation” exist, forms that “can even be realised with workers’ active cooperation and possess an appearance of legality.” Workers sign contracts, accept terms, and click “agree”—formal voluntariness precisely constitutes substantive unfreedom.
六、批判与反思：奴役理论的边界 / 6. Critique and Reflection: The Boundaries of Slavery Theory
当代奴役理论虽然提供了强大的批判工具，但也面临不容回避的理论挑战。
Although contemporary slavery theory provides powerful critical tools, it also faces theoretical challenges that cannot be avoided.
首先是概念泛化的风险。如果一切形式的劳动控制都可以被命名为“奴役”，那么这个概念本身就失去了分析力。新罗马主义的工资奴役理论已经意识到了这一问题，并试图通过“意志屈从”和“基本需求依赖”这两个限定条件来约束其适用范围。但数字奴役理论是否也需要类似的约束机制，仍是一个未解决的问题。
The first is the risk of conceptual overgeneralisation. If all forms of labour control can be named “slavery,” then the concept itself loses analytical power. Neo-Roman wage slavery theory has already recognised this problem and attempts to constrain its scope through two qualifications: “subjection of the will” and “dependence on basic needs.” But whether digital slavery theory also needs a similar constraining mechanism remains an unresolved question.
其次是西方中心主义的陷阱。有学者对“现代奴役—气候变化”这一新兴议题提出了尖锐批评，指出这一叙事实际上“复活了早已被否定的环境决定论”，其效果是为西方开脱其对劳动剥削和环境破坏的历史责任，“同时将行动负担转嫁给全球大多数”。当代奴役理论如果不能反思自身的话语政治，就可能沦为一种新的“拯救主义”——西方学者为全球南方的“被奴役者”代言，却回避了自身在维持不平等结构中的位置。
The second is the trap of Western-centrism. Some scholars have sharply criticised the emerging “modern slavery–climate change” nexus, pointing out that this narrative actually “revives the long-discredited environmental determinism,” and its effect is to absolve the West of historical responsibility for labour exploitation and environmental destruction “while shifting the burden of action onto the global majority.” If contemporary slavery theory cannot reflect on its own discursive politics, it may become a new form of “saviourism”—Western scholars speaking for the “enslaved” of the Global South while avoiding their own position in maintaining structures of inequality.
第三是行动路径的缺失。技术封建主义理论擅长诊断，却未必擅长开方。瓦鲁法基斯等人主张用“数字公有化”来对抗平台垄断，但这一方案在具体制度设计上仍显模糊。相比之下，数字劳动研究中提出的“平台合作主义”、数字劳动的法律承认和“普遍数字收入”等方案，虽然更为具体，但其可行性和有效性仍有待实践检验。
The third is the absence of an action path. Techno-feudalism theory is good at diagnosis but not necessarily at prescription. Varoufakis and others advocate “digital public ownership” to counter platform monopoly, but this proposal remains vague in concrete institutional design. By contrast, proposals in digital labour studies—such as “platform cooperativism,” legal recognition of digital labour, and “universal digital income”—are more specific, but their feasibility and effectiveness still await practical testing.
七、结论 / 7. Conclusion
2025—2026年的奴役理论发展呈现出一个清晰的学术脉络：从工业资本主义的工资奴役，到数字平台经济的数字奴役，再到技术封建主义的“云农奴”处境，理论界正在构建一条奴役形态的演进链。这条演进链的核心发现是——奴役的形态在变化，但奴役的本质未变。
The development of slavery theory in 2025–2026 presents a clear academic trajectory: from wage slavery in industrial capitalism, to digital slavery in the platform economy, and further to the “cloud serf” condition under techno-feudalism. Theorists are constructing an evolutionary chain of slavery forms. The core finding of this chain is that the forms of slavery are changing, but the essence of slavery has not changed.
变化的，是从暴力强制到算法控制、从人身占有到意志屈从、从公开的锁链到隐形的数据枷锁。不变的，是一部分人的自主性以另一部分人的利益为代价被系统性地削减。
What has changed is the shift from violent coercion to algorithmic control, from physical ownership to subjection of the will, and from visible chains to invisible data shackles. What has not changed is that the autonomy of some is systematically reduced for the benefit of others.
当代奴役理论的最大价值或许不在于它给出了什么答案，而在于它迫使我们面对一个不舒服的问题：当我们自认为生活在一个“自由”的时代，我们是否只是在用更精致的方式，延续着人类最古老的不平等？
Perhaps the greatest value of contemporary slavery theory lies not in the answers it provides, but in the uncomfortable question it forces us to confront: when we believe we are living in an age of “freedom,” are we merely continuing humanity’s oldest inequality in a more refined way?
参考文献 / References
[1] Bales, K., & Paz-Fuchs, A. (2026). Unfree Labour. In Underpaid and Unrecognized: State-Induced Unfree Labour. Oxford University Press.
[2] Ruggiu, D., & Özdemir, B. (2026). Behind the AI’s Gaze: Generative Artificial Intelligence, Microworkers and Digital Slavery. NanoEthics, 20(1), 2.
[3] Wage Slavery: A Neo-Roman Account. (2026). Political Studies (online first, April 2026).
[4] Arvidsson, A., & Bonini, T. (2026). Factories or Workhouses? Digital Platforms and the Re-Feudalisation Thesis. SAGE Journals.
[5] Rethinking AI’s “Liberation” Fantasy: Free Us? Or Chain Us Further? (2025). AI & Society, 40, 6507–6508.
[6] Bhagat, A., & Kenis, A. (2026). The Modern Slavery–Climate Change Nexus. Antipode, 58(1).
[7] Gabelaia, I. (2026). The Relevance of Algorithmic Control and AI in Employee Onboarding in the Corporate Slavery Context. In Entrepreneurship and Human-Centric Business Strategies. Springer.