The development of artificial intelligence technology will challenge the governance of human society in the future，The extensive embedding and reconstruction of the social technology system in the field of public survival，This involving the new type of distribution order between the autonomy of intelligent machines and human control。In recent years，With autonomous driving、Surgical Robot、Smart Home、Smart Recruitment、The commercialization and social application of the algorithm decision -making system of intelligent consultation and other algorithm decision -making system，Make human subjects constantly transfer control in the key decision -making field，while the autonomy of the intelligent machine expands，The ethics of artificial intelligence is added accordingly、Questions of legal and social supervision。In this situation，Artificial intelligence governance re -emphasizes that humans must ensure the control of the independent decision -making behavior of intelligent machines。But in the sense of direct operation today，How to define more abstract control relationships，How to maintain moral norms and clear human responsibilities? This，With "meaningful human control" (Meaningful Human Control，MHC) This emerging artificial intelligence ethics governance principle and its theoretical framework to guide practice。

　　bet365 best casino games

In recent years，Responsible innovation、Value sensitivity design、Ethical design concepts of value embedding in the research and development phase emphasized in research and development stages of moral materialization，Occupied the mainstream of the global governance field of artificial intelligence。but，Even in the early stage of technology research and development, value bet365 Play online games embedded，It may still be unable to avoid potential risks caused by artificial intelligence when applying。The main reason is the following two points: on the one hand，Artificial intelligence is a transformative technology，Can fully penetrate human survival space and reshape social production relations。So，The value concept of guiding artificial intelligence applications is dynamic evolution。On the other hand，The iterative speed of artificial intelligence technology is very fast，Especially in terms of gradually replacing the control of human decision -making, it has a profound impact。In the foreseeable future，"Completely Autonomous" smart machine systems may appear in the field of artificial intelligence，or bring significant challenges to human social order。In traffic、Medical、Home Furnishing is equivalent to human life closely related fields，Sports to the full decision -making of the smart machine，It may not meet human rights and morals。So，Human and institutions instead of artificial technology and algorithms，Decisions of the high -level autonomous system should be finally controlled，And assume ethical and legal liability。

This framework assumption premise，The earliest system of artificial intelligence -driven for human survival- "Lethal Autonomous Weapon Systems，LAWS) Field。2012，The US Department of Defense issued 3000.09 instructions，For the first time, the whole life cycle of the autonomous weapon system is proposed at other levels of the country must be controlled and fully reflects the human commander、The intention of the operator。2014，The International Committee of the Red Cross conveys the concept of MHC on the key functions of LAWS to LAWS，formed a wide range of consensus。In the same year，The United Nations Convention on Specific Customs Weapon officially began to include LAWS follow MHC's ethical issues into the framework agenda。until the current，bet365 best casino games MHC has developed into the ethics of LAWS in the international community、The core of the concept of legal and public policy framework。but but，This concept still lacks a unified definition at the practical level。The main reason is，LAWS has special characteristics，It is not a social technology system that is generally related to the existence of human survival and life in the future，It is unlikely to subvert the traditional public life mode。But the study of the concept of MHC from this field，It can reveal the basic conditions of humans control intelligent machines and assume moral responsibility: make conscious decisions、Fully understand the rationality of the behavior and its situation、Master the skills and knowledge of the application autonomous system。

but but，These conditions are not easy to achieve。The fundamental ethical challenge of artificial intelligence is the problem of liability gap，This is mainly to gradually appear during the human -computer interaction of complex scenes，Lead technical、Economics、Morality、Mixed intervention of various reasons such as political nature。Another，Responsibility Honggou also has many types，Not just moral responsibility，There is also responsible for being prepared、Public office responsibility、Active responsibility, etc.。The interaction between people and high -level smart machines，Generally not in the sense of direct operation，cannot be simply defined as "one -on -one" (the operator directly guides the behavior of technical equipment or system) causal responsibility。Therefore，Between human and intelligent machines is a novel "distributed control" relationship，The basis for action based on the attribution of responsibility，New theoretical frameworks are needed。

MHC concept has two basic compositions: "meaningful" and "control"。"meaningful" refers to the basic condition that replaces the behavior of humans to control smart machines and assume moral responsibility。"Control" bet365 Play online games involves the diversification of the control form of the human subject's control of intelligent machines。Collection of different forms of control and responsibility，It is conducive to the realization of the supervision mechanism of intelligent machine behavior。This is the problem of dissolving the responsibility of the gap，Provided theoretical entry point。Filippo Santoni de SiO and Jeroen Van Den Hoven conducted a pioneering demonstration on MHC's moral philosophy foundation。Their argumentation and cognitive science based on their contemporary contemporary theory、The latest research results of spiritual philosophy，Provided a regular framework for the dual conditions of "tracking -traceability" under the perspective of social technology system。

The principle of "tracking" involves the relationship between the human controller and the control system，That is, systematic behavior can reflect the reasons for human action in the practical situation anytime, anywhere (intention、plan or moral rationality) extent or quality。The concept of this concept in philosophy and ethics，Allows the control form of non -direct operating sense。"Tracking" is a supplement to "tracking"，Refers to the behavior of the system should be traced back at least one responsible agent (user、Supervisor、Designer、Legislator, etc.)。These agents must master the explicit knowledge and the knowledge of the technical，and realize the identity of your own moral suffixer。

　　Ethical Righteousness Mechanism 

Current，MHC theory has stipulated the general content and goals。Based on the dual design principle of "tracking -trace"，MHC -oriented intelligent machine ethics design，It is based on the overall social technology system，Not limited to a single technical hardware and the value embedded in its research phase。but but，There are some dilemma to convert the specifications of MHC into feasible operations: the tracking standard needs to determine who Bet365 lotto review the agents are、What are the reasons for action; traceability standards need to identify whether these agents have both technical expertise and moral consciousness。

Operationable dilemma for tracking principles，Giulio MECACCI and others based on action philosophy、Research results of behavioral psychology and traffic engineering，I proposed a connection、"Proximal Scale of Reason"。Operating standards for tracking principles，Need an agent、A degree of influence on system behavior can be quantified，Define the subject of morality and legal liability with this。Agent and system at time、The proximal and remote end on the space is the basis for dividing，The agent can be divided into the internal agent of the system and the external agent of the system，Then the distinction between the characteristics of the characteristics of the internal and external systems can be distinguished by the distant and near internal and external systems。

Realization of the principle of traceability，Need to measure how much human beings understand the design of smart machines from the level of technical expertise and moral consciousness、deployment and operation，is closely stable on the control chain。But there are difficulty in quantifying the abstract element here。Simeon Calvert has developed a "Evaluation Cascade Table"，Can be related to knowledge、Pure qualitative assessment of ability and intention provides an empirical basis。

First，Need to refine the type of technical expertise and moral consciousness in the control system。For the intelligent system，The technical expertise of the agent can be divided into systemic knowledge and decision ability。The former involved the knowledge of (formalization and informal) knowledge of intelligent machines, as well as existing (active or passive) experience，The latter is the reason for implementation action (strategy、Tactics、Operation) Ability。Moral consciousness includes basic value standards (accountability、Transparency、bet365 best casino games Explanable, etc.) understanding and moral ownership awareness (ethics and legal liability、Positioning on the impact on others and your own responsibilities)。

Next，Knowledge、The ability to control the system control at the level of the sense of morality，Use an evaluation level to conduct perception survey。This is a comprehensive application based on the "LIKERT Scale"，Operation control from step by step、Participation of human agents、Agent's ability to understand and use the system、The agent understands their ability to make the system's moral responsibility，In these four aspects，Determine the degree of participation through the addition of total scores。

The key to the next research is，The integration of the "tracking -trace" dual standards in the social technology system becomes a problem。How to judge the technical expertise and the agent of moral consciousness who meet the traceable conditions and the agent who realizes the tracking conditions is the same subject，and fulfill the corresponding laws and regulations，All need to further expand the philosophical framework and case study of MHC。

Overall，MHC as the meaning of ethical principles is to reiterate the right to human security in the field of artificial intelligence，Its theoretical purpose and practical goal are to solve the problem of technology control，to fully reflect human reasons and intentions。MHC Study rapid advance，It indicates that artificial intelligence global governance enters a new stage。The operator of the intelligent system may no longer be regarded as the main subject of accountability，The degree of participation on the agent on the intelligent system control chain will determine its responsibility position。my country is in a period of opportunity for artificial intelligence to flourish，Research on artificial intelligence ethics at the forefront of the international community，Timely reference and improve the application of local policies and regulations。In MHC's moral philosophy bet365 live casino games framework，Ethical elements and accountability mechanisms that can be included in the local area to expand the theoretical foundation，Then guide、Supervise the full life development cycle of artificial intelligence。

　　(This article is the major project of the National Social Science Foundation "In -depth Promoting the Reform of the Science and Technology System and Improve the Study of the National Science and Technology Governance System" (21ZDA017) 

(Author Unit: School of Social Sciences, Tsinghua University)