The Ethical Dimension of AI Governance in the New Era

General Secretary Xi Jinping, when presiding over the twentieth collective study session of the Political Bureau of the CPC Central Committee, stressed that “we must comprehensively advance AI technological innovation, industrial development, and enabling applications, improve the AI regulatory system and mechanisms, and firmly grasp the initiative in AI development and governance.” In the era of artificial intelligence, the issue of AI governance has transcended the boundaries of the technical sphere and become a core global governance topic affecting the future development of humanity. From the very inception of AI, its creation, application, and its very nature as a disruptive technology have been accompanied by ethical concerns, posing a series of new challenges to traditional frameworks of moral governance. In the actual situation where AI empowers every industry and sector, it has also given rise to a demand for comprehensive and forward-looking ethical governance. It can be said that the construction of AI ethics and its governance pathways is not only a phased task of technical governance but also an inherent requirement for delineating the ethical boundaries of a future digital-intelligent society.

习近平总书记在主持中共中央政治局第二十次集体学习时强调，“全面推进人工智能科技创新、产业发展和赋能应用，完善人工智能监管体制机制，牢牢掌握人工智能发展和治理主动权”。人工智能时代，人工智能治理问题已超越技术范畴的界限，成为牵动人类未来发展的全球治理核心议题。从人工智能产生伊始，其制造、应用及自身作为一种颠覆性技术就伴生着关于伦理的隐忧，使传统道德治理框架面临一系列新挑战，在人工智能赋能千行百业的现实境况中，也催生了全域性、前瞻性的伦理治理需求。可以说，人工智能伦理及其治理路径的建构不仅是一项技术治理的阶段性任务，更是未来数智社会伦理边界厘定的题中应有之义。

Constructing Ethical Norms for AI Governance

建构人工智能治理的伦理规范

The core objective of constructing ethical norms for AI is to embed moral judgments and action principles into intelligent systems that conform to the laws of human social development, ensuring that throughout the entire process of empowering economic and social development, they always take the safeguarding of humanity’s overall interests as their fundamental guideline. In essence, among the many fields of applied ethics, no other field is as closely linked to the concept of responsibility as the ethics of science and technology. To realize this vision, the field of ethics has undergone a theoretical evolution from Asimov’s “Three Laws of Robotics,” to Wiener’s “machines should serve human needs,” and then to the “value alignment” advocated by Myner and others. These studies have not only pointed out the difficulties of human-machine value alignment but also highlighted the crucial role of moving from “value alignment” to “value symbiosis.” However, due to the spatiotemporal limitations of human cognition, the inherent difficulties of value encoding, and the technical boundaries of algorithm operation, AI ethical theory faces significant operational limitations. Undoubtedly, the construction of AI ethical norms cannot stop at aligning machines with static human values; it should proactively move toward a new stage of collaborative development in which humans and machines achieve mutual understanding, mutual adjustment, and mutual evolution through interaction.

建构人工智能伦理规范的核心目标是为智能系统嵌入符合人类社会发展规律的道德判断与行动准则，确保其在赋能经济社会发展的全过程中始终以维护人类整体利益为根本遵循。实质上，在应用伦理学的诸多领域中，没有任何一个领域像科技伦理那样同责任概念联系如此紧密。为了实现这一愿景，伦理学界经历了从阿西莫夫的“机器人三定律”，到维纳的“机器应服务人类需求”，再到迈内尔等人倡导的“价值对齐”的理论演进。这些研究不仅指明了人机价值对齐的难点，更凸显了从“价值对齐”迈向“价值共生”的关键作用。然而，由于人类认知的时空局限、价值编码的固有困难及算法运行的技术边界，人工智能伦理理论面临显著的运行局限。毫无疑问，人工智能伦理规范的建构不能止步于让机器对齐静态的人类价值，而应前瞻性地迈向人与机器在互动中实现共同理解、共同调适与共同演进的协同发展新阶段。

How can we, against the backdrop of an era characterized by technical complexity, algorithmic opacity, and diversity of values, explore a set of embeddable, verifiable, and sustainable ethical norms for AI? This has become the core of the issue. This requires relevant responsible subjects to transcend purely technical boundaries and construct a systematic project that requires coordinated advancement in technological breakthroughs, philosophical reflection, institutional innovation, and social dialogue. Since 2024, the intensive release of global documents such as the *Beijing AI Safety International Consensus*, the *Shanghai Declaration on Global AI Governance*, and the *Tianjin Declaration of the Council of Heads of State of the Shanghai Cooperation Organization* marks the formation of key consensus within the international community on AI ethical identification. These documents aim to jointly construct an ethical governance framework for the AI era, centered on human rights protection, social justice, and human dignity. However, the implementation of this consensus faces deep challenges. Traditional notions of “value alignment” are deeply entrenched, whereas value coordination in reality is actually a dynamic, two-way process of adjustment and symbiosis. In view of this, the construction of AI ethical norms at the metaphysical level must trace back the questions of the times: how AI ethics takes shape, how it establishes its foundation, how it overcomes difficulties, and how it adapts to change. These inquiries concerning AI ethics constitute the inherent “value compass” of AI entities, making it possible for them to truly become trustworthy civilizational partners for humanity, thereby building a future of human-machine co-prosperity and intelligent goodness.

我们如何在技术复杂性、算法不透明性与价值观念多样性的时代背景下，探索出一套可嵌入、可验证、可持续的人工智能伦理规范成为问题的核心。这要求相关责任主体超越单纯的技术边界，构建一个需要技术突破、哲学反思、制度创新与社会对话协同推进的系统工程。2024年以来，《北京AI安全国际共识》《人工智能全球治理上海宣言》《上海合作组织成员国元首理事会天津宣言》等全球性文件的密集出台，标志着国际社会在人工智能伦理认同上形成关键共识，意在以人权保障、社会公正与人格尊严为核心，共同构筑人工智能时代的伦理治理框架。然而，共识的落地面临深层挑战，传统“价值对齐”观念根深蒂固，而现实中的价值协同实为动态双向的调适共生。有鉴于此，形而上层面的人工智能伦理规范的建构必须追溯人工智能伦理何以成形、回应人工智能伦理何以立本、破解人工智能伦理何以克难、架构人工智能伦理何以应变的时代之问。这些关涉人工智能伦理的追问构成人工智能体内在的“价值罗盘”，使其有可能真正成为人类可信赖的文明伙伴，进而构筑人机共荣、智能向善的未来。

In the practical construction of AI ethics, there are three possibilities: top-down norm implementation, bottom-up experiential learning, and a hybrid of the two. These three possibilities for constructing AI ethics not only shape the development direction of AI ethical theory but also profoundly influence AI ethical practice on a global scale. Top-down norm implementation involves encoding existing human ethical principles into computational symbols and embedding them into AI systems through algorithmic instructions, thereby achieving the goal of regulating the behavior of AI entities. Bottom-up experiential learning involves simulating evolution and moral acquisition. Through machine learning and the evolution of system organization, AI induces moral principles from specific situations, forms decision-making capabilities, and ultimately completes an autonomous evolution from experience to norms. The hybrid approach integrates top-down and bottom-up methods. Based on the guidance of ethical theory, it dynamically generates and optimizes ethical norms through simulation, training, and trial and error, while simultaneously incorporating functions such as emotion and cognition to enhance moral decision-making capabilities. It also emphasizes interdisciplinary collaborative research involving philosophy, computer science, and other fields.

在人工智能伦理的实践建构中，有三种可能性，即自上而下的规范贯彻、自下而上的经验学习以及融合二者的混合。这三种人工智能伦理构建的可能性不仅塑造了人工智能伦理理论的发展方向，也深刻影响着全球范围内的人工智能伦理实践。自上而下的规范贯彻是将人类既有的伦理原则编码为计算符号，通过算法指令嵌入人工智能系统，达到规范人工智能体行为的目的。自下而上的经验学习是通过模拟进化与道德习得，借助机器学习与系统组织演化，使人工智能从具体情境中归纳道德原则、形成决策能力，最终完成从经验到规范的自主演进。混合进路是综合自上而下与自下而上的方法，基于伦理理论指导，通过模拟、训练与试错动态生成并优化伦理准则，同时融入情感、认知等功能来增强道德决策能力，并强调哲学、计算机科学等多学科协同研究。

The Ethical Essence of AI Governance

人工智能治理的伦理本质

The increasingly approaching digital-intelligent era has already made AI and its ethical issues no longer merely a matter of metaphysical philosophical speculation, but an increasingly urgent problem of applied ethics. As a disruptive technology, AI inevitably generates complex ethical controversies and governance challenges in the process of its technological evolution and social application.

日益向我们走来的数智时代已经使人工智能及其伦理问题不再仅仅是形而上层面的哲学思辨，而是日益紧迫的应用伦理问题。人工智能作为一种颠覆性技术，在技术演进与社会化应用的过程中，必然产生复杂的伦理争议与治理难题。

Looking back at the mainstream theoretical pathways and governance practices of AI ethical construction, it is not difficult to identify three essential issues that run through them all and are interrelated: First, the construction of AI ethics blurs the boundary between two fundamentally different behavioral motivations: “altruism based on self-interest” and “pure altruism.” In the age of AI, information capitalism has exacerbated data monopolies. A minority of subjects holding information advantages engage in the abuse of power, evading responsibility under conditions of imbalanced rights and responsibilities, shifting technical risks and catastrophic consequences onto the public, ultimately forming a systemic risk where a minority profits while the majority bears the costs. Second, the construction of AI ethics has failed to guide the dynamic evolution of cultural concepts within technological development, and has not transformed the most sublime spiritual values of human civilization into principles that machines can understand and follow. Finally, the process of constructing AI ethics has failed to fully identify and respond to the fundamental and urgent threats facing human existence, falling into collective unconsciousness and historical nihilism. In other words, AI’s dissolution of the human spiritual subjectivity will lead to the blurring of the meaning of human existence and the hollowing out of the sense of value, causing people to lose their exploration of the meaning of life.

回望人工智能伦理建构的主流理论路径与治理实践，不难辨识出三个贯穿始终、相互关联的本质问题：首先，人工智能伦理建构模糊了“以利己为出发点的利他”与“纯粹的利他”这两种根本不同的行为动机之间的界限。在人工智能时代，信息资本主义加剧了数据垄断。少数掌握信息优势的主体存在滥用权力的行为，在权责不对等的情况下规避责任，将技术风险与灾难性后果转嫁给公众，最终形成少数人获利、多数人承担代价的系统性风险。其次，人工智能伦理建构未能引导技术演化中文化观念的动态流变，未将人类文明中最崇高的精神价值转化为机器可理解、可遵循的内在原则。最后，人工智能伦理建构的过程未能充分辨识并响应人类生存所面临的根本性、紧迫性威胁，陷入集体无意识和历史虚无主义。换言之，人工智能对人类精神主体性的消解，将导致人的存在意义模糊与价值感虚化，使人失去对生命意义的探索。

Existing research, when responding to AI ethical risks, has intentionally or unintentionally evaded two fundamental issues: On the one hand, when AI is fully integrated into the processes of human production and life, it is both regulated by the internal logic of scientific exploration and externally shaped by capital accumulation and social power. How to handle this contradictory yet unified opposition is an essential issue that the construction of AI ethics cannot avoid. The fundamental driving force of technological development—whether it originates from scientific autonomy or is constrained by the shaping of capital and power logic—directly determines the foundation of AI ethical construction. The structural “labor substitution” and “capacity deprivation” brought about by scientific and technological progress reshape the traditional value of labor and the foundation of existence; the technical alienation of intelligent design subjectivity makes the monopolization of a minority of interest subjects possible, while the interests of the majority fall into functional incapacitation due to technical exclusion. The public’s increasing dependence on intelligent systems, and the cognitive gap, continuously erode human autonomy and dignity. On the other hand, when human society falls into a value dilemma and moral challenges due to the excessive application of AI, whether we can construct ethical norms for AI becomes another essential problem urgently needing a solution. In short, do ethical rules in the technical field become alienated into some form of human-machine “rules of the game”? If the construction of AI ethics remains only at the level of technical risk control, ignoring the profound logic of capital regulation and structural inequality behind it, governance paradoxes such as “using algorithms to prevent cheating” will continue to recur.

现有研究在回应人工智能伦理风险时，有意无意地回避了两个根本性问题：一方面，当人工智能全面融入人类生产生活过程，其既受科学探索内在逻辑的规制，亦受资本增殖与社会权力的外在塑造，而如何处理这个既对立又统一的矛盾是人工智能伦理建构无法回避的本质性问题。技术发展的根本驱动力——究竟源于科学自主性还是受制于资本与权力逻辑的塑造，直接决定了人工智能伦理建构的基点。科学技术进步带来的结构性“劳动替代”与“能力剥夺”重塑了传统劳动价值与生存根基；智能设计主体性的技术异化让少数利益主体的垄断性成为可能，但多数人的利益却因技术排斥陷入功能性失能。公众对智能系统的日益依赖，认知鸿沟持续消解着人的自主性与尊严。另一方面，当人类社会因人工智能过度应用而陷入价值困境与道德挑战时，我们能否构建起人工智能伦理准则，成为另一个亟待破解的本质问题。简言之，技术领域的伦理规则是否异化为某种形式的人机“游戏规则”？若人工智能伦理建构仅停留在技术层面的风险管控，视其背后深刻的资本逻辑规制与结构性不平等于不顾，“用算法防作弊”之类的治理悖论将不断重现。

Overall, the process of resolving the difficult problems of AI ethical construction is also a process of examining human “ontology” and “coexistence theory,” and even more so a deep awareness of and active response to the spiritual world and existential predicament of humanity. Effective AI ethical construction must take into account both the critical examination of technological tools and the continuous renewal of human value practices. Only when AI entities manifest altruistic behavior in ethical design can the reflexive governance process of generative AI truly commence.

总体而言，破解人工智能伦理建构的疑难问题的过程也是审视人类“存在论”与“共存论”的过程，更是对人类精神世界与生存困境的深度觉察与主动回应。有效的人工智能伦理建构必须兼顾对技术工具的批判性审视与人类自身价值实践的持续革新。只有人工智能体在伦理设计中表现为利他行为，自反性的生成式人工智能治理过程才能真正开启。

The ethical innovation of AI governance

人工智能治理的伦理创新

The process of AI ethical construction is essentially a process in which humanity actively shapes the ethical direction of scientific and technological evolution through moral reason. Specifically, resolving the difficult problems of AI ethics requires innovative measures that are compatible with them. This can be achieved by advocating the principle of shared risk to address ethical risks and form an “internal symbiosis” structure. At the same time, adhering to the perspective of “coexistence theory” to respond to human existential crises, thereby realizing a transformation of AI ethical construction from passive adaptation to active co-construction.

人工智能伦理建构的过程实质上是人类凭借道德理性主动塑造科学技术演进的伦理方向的过程。具体而言，破解人工智能伦理疑难问题需要与之相适配的创新举措，可以通过提倡风险共担原则应对伦理风险形成“内共生”结构。同时，秉承“共存论”视角回应人类生存危难，进而实现人工智能伦理建构从被动适应到主动共构的转变。

First, advocate the principle of “shared risk” to address ethical risks. The current application of AI technology merely synthesizes, simulates, and transforms specific functions of the human brain. At this stage, robot designers and manufacturers are essentially antecedent ethical agents and must jointly bear human-machine risks, such as information asymmetry, power-responsibility asymmetry, and cognitive asymmetry, to avoid the phenomenon of risk transfer. Second, form an “internal symbiosis” structure to guide meme mutation. In responding to the challenge of meme mutation, humanity and AI technology must construct a tightly interdependent symbiotic system. In this system, the two are not a simple subject-object relationship but a continuously interacting collaborative relationship, jointly forming a dynamically adjusted “internal symbiosis” ecosystem. This symbiotic relationship can not only change people’s ways of life but also help promote the progress of human society. Finally, adhere to the perspective of “coexistence theory” to respond to human existential crises. In the process of AI ethical construction, the coexistence of responsibility and risk must be taken as the highest value goal, systematically preventing and responding to those extreme, existential risks that may endanger the overall survival of humanity. In essence, to share moral responsibility, the behavioral agents of AI—from large entities such as nations and ethnic groups down to designers and manufacturers—must, in some sense, control their actions and consider their moral responsibility as a whole.

首先，提倡“风险共担”原则应对伦理风险。现今人工智能技术的应用仅是将人类大脑的特定功能合成、模拟以及转化出来。在这个阶段，机器人设计者、制造商实质上为前置的伦理行为主体，必须共同承担人机风险，如信息不对称、权责不对称、认知不对称等，避免产生风险转移现象。其次，形成“内共生”结构引导模因突变。在应对模因突变的挑战时，人类与人工智能技术必须建构一种紧密依存的共生系统。在此系统中，二者并非简单的主客关系，而是持续互动的协同关系，共同构成一个动态调整的“内共生”生态系统。这种共生关系不仅可以改变人们的生活方式，而且有助于推动人类社会的进步。最后，秉承“共存论”视角回应人类生存危难。在人工智能伦理建构的过程中，必须将责任与风险共存作为最高价值目标，系统性地预防与应对那些可能危及人类整体存续的极端性、生存性风险。实质上，为了分担道德责任，人工智能的行为主体大至国家、民族，小至设计制造者都必须在某种意义上控制其行为并在整体上考量其道德责任。

In an era when AI is highly concerned and also generates considerable controversy, we must calmly consider the ethical issues in AI governance, rationally plan the development direction of AI, and put the “ethical” reins on it. With the profound transformation of AI technology, the mode of human existence and moral practice is undergoing a shift from stable subjectivity to fluid symbiosis. The traditional status of humanity as the sole center of cognition and action is gradually dissolving, individual autonomy and dignity face profound challenges, and the foundations of social fairness and justice are being eroded. Faced with the systemic ethical risks brought by AI, there is an urgent need for an emergency ethics that can shape value coordinates, forge a moral foundation, and construct an ethical order. In the future society, whether it is human-machine collaboration or human-machine symbiosis, it is necessary to highlight the brilliance of humanity, deeply insight into the future development direction of humanity, actively prevent difficult problems in the development of human society, and guide the development of an intelligent society onto a healthier and more civilized track. Thus, by resolving the difficult problems of AI ethical construction and its governance, and by cultivating a profound sense of crisis and value resilience, we can build a dynamically stable and trustworthy ethical baseline for the future of human-machine symbiosis.

在人工智能受到高度关注，同时衍生较大争议的时代背景下，我们必须冷静地思考人工智能治理中的伦理问题，合理规划人工智能的发展方向，为其套上“伦理”规则的缰绳。随着人工智能技术的深刻变革，人类的存在方式和道德实践正经历从稳固的主体性向流动的共生性转变。人类作为认知与行动唯一中心的传统地位逐渐消解，个体的自主性与尊严面临深刻挑战，社会公平和正义的基石受到侵蚀。面对人工智能带来的系统性伦理风险，迫切需要一种能塑造价值坐标、铸就道德基底、建构伦理秩序的应急伦理。未来社会，无论是人机协同，还是人机共生，都需要彰显人性光辉，深刻洞察人类未来发展方向，积极防范人类社会发展疑难问题，引导智能社会的发展步入一个更为健康文明的轨道。由此，破解人工智能伦理建构及其治理难题，通过培育深刻的危机意识与价值韧性，为人机共生的未来筑牢动态稳定、值得信赖的伦理基线。

Whether viewed from the trend of AI leading a paradigm shift in scientific research or from the needs of the modernization of the national governance system and governance capacity, AI governance must be considered within the perspective of ethics. Through an effective elucidation of AI ethical construction, we can avoid stopping at aligning machines with static human value deductions; through an analysis and grasp of the essence and boundaries of AI ethics, we can always maintain a sober attitude toward AI and its governance; according to the actual development of AI, we must continuously enrich and innovate ethical theory, understand, accommodate, and guide AI development, so that the discipline of ethics keeps pace with scientific and technological innovation. Only by organically combining ethical construction, ethical governance, and ethical innovation can we promote the upward and good development of AI, forming a win-win situation of healthy technological development and continuous social progress.

无论从人工智能引领科研范式变革的趋势来看，还是从国家治理体系和治理能力现代化的需求来看，人工智能治理都必须在伦理学的视域中加以考量。通过对人工智能伦理构建的有效阐明，避免止步于让机器对齐静态的人类价值推演；通过对人工智能伦理本质及其边界的分析和把握，始终保持对人工智能及其治理的清醒态度；根据人工智能的发展实际，不断丰富和创新伦理学理论，理解、包容和指导人工智能发展，使伦理学学科发展紧跟科技创新步伐。唯有将伦理构建、伦理治理与伦理创新有机结合起来，才能促进人工智能向上向善发展，形成科技健康发展与社会持续进步的双赢局面。

(Author: Lu Xiangfeng, Professor at the School of Marxism, Tianjin Medical University)

（作者：路向峰，系天津医科大学马克思主义学院教授）