Keywords

1 Overview and Category of Artificial Intelligence

“Artificial intelligence’’ was born in an academic conference in 1956 at Dartmouth in the US. Several outstanding young scientists gathered to discuss the issues of intelligence, who did their research on mathematics, neuroscience, computer science, psychology and many other aspects. The concept of Artificial Intelligence was proposed by John McCarthy in the conference. The main goal of artificial intelligence is attempted to achieve human intelligence on computers and research results have emerged in endlessly from the beginning. Every achievement of artificial intelligence has different features and functions. For example, Siri focuses on speech recognition, AlphaGo focuses on machine learning, expert systems focus on inference with domain knowledge, and so on. The results of AI have different characteristics and functions, and can be broadly classified into three categories in terms of the relationship between the external behavior of AI and innate design and learning.

The first type of AI whose outwardly manifested behavior is only related to its innate design and is not capable of acquired learning. The skills or functions possessed by this type of AI are unique, and no amount of later design can change its innate functions. There are many AI products in this category, such as smartphones, computers, cars, etc. The characteristic of this class of AI is that its ability behavior is only influenced by the innate design, and the acquired learning and training are meaningless.

The second type of artificial intelligence is not only related to innate design, but also this type of AI is capable of machine learning. This type of AI is initially set to have certain functional characteristics and has some prototype of machine intelligence, and only after a later learning and training process does it slowly master this skill or perform certain operations or behaviors. For example, AlphaGo was designed by Google to have the initial skills of Go, and it was only after a lot of data training or machine learning that AlphaGo was really able to master the skills of Go. AlphaGo is a good illustration of this type of AI, which initially has a clear function, and through machine learning and other means, its functional characteristics tend to be perfected and eventually reach a stable state, thus completing machine learning. The functions or behaviors of this class of AI are subject to both limitations from initial design and influences from later learning. In this class of AI, the initial design plays a crucial role in determining that the AI actually has a certain skill, as well as in determining the possibility and significance of later learning training.

The third category of AI is that which also has a predetermined design of an intelligent system that is also capable of learning, but this learning is not limited to a particular domain and can be broad and non-domain specific. The predetermined design of such an intelligent system is only a framework for thinking; it is the training and learning that are more fundamental factors in shaping this type of AI. The first thing that distinguishes this type of AI system from the second type of AI system is that the innate design of the system does not have a clear and specific functional orientation; such systems are only responsible for thinking, reasoning, computing, decision-making, etc., and are open to the tasks and needs of the external world. Specific functions and operations need to be determined according to the problems and goals in practice, and are also adaptive to the external world, so this type of AI does not have a particular single or closed function, and shapes its capabilities based on learning and experience. The typical representative of this type of AI is the NARS system designed by Prof. Pei Wang of Temple University, and also some artificial general intelligence systems. The most important feature of NARS is adaptability, that is, NARS is able to respond to future situations with limited knowledge and relatively insufficient resources, and respond to unlimited needs with limited resources.

The relationship of the function and characteristics of these three types of artificial intelligence between design and learning can be shown in the figure (Fig. 1) .

Fig. 1.
figure 1

Category of artificial intelligence

Full size image

Because the acquired learning training is related to the time factor, the longer the time, the more adequate the machine learning will be. Therefore, the functions and capabilities of AI can be represented in the following graphical way. The first type of AI is also the one whose capabilities depend only on pre-design, so it is represented as a straight line, such as a smartphone, which has only those functions. The second type of AI is influenced by both innate design and acquired learning, but their functional capabilities are closed and through training will also converge to a certain state of perfection in the end, so the second type of AI presents as a curve with a gradual slope towards 0, such as AlphaGo. The third type of AI is developed for the external world because it has only one learning frame of mind innately, and the acquired learning makes it possible to have different functions and capabilities. Thus, as learning continues to strengthen over time, the capabilities of the third AI are also being strengthened. Thus, it appears roughly as a straight line with a constant slope. These three categories are also categorized to be called as the first one is no intelligence, the second one is professional AI, and the third one is AGI, or artificial general intelligence (AGI), and it can be seen from the image that intelligence is actually their slope. AGI because its goal is to achieve human intelligence, and the level of human intelligence is an invariant, so the intelligence of AGI is also presented as a straight line. There is no possibility of a curve with an increasing slope, and that would be what the media call super AI.

2 The Influence of Artificial Intelligence on the Law

These three types of AI agents have had or are having a profound impact on our society, and the impact of the first two types of AI has been much discussed and researched. However, the three types of AI that have the most profound impact on our society and legal system have been relatively little studied. This paper focuses on the impact of the third type of AI, i.e., general-purpose AI.

Under the current legal framework, there are only two possible types of legal regulation of artificial intelligence in legal relationships: one is as an object, the object of direct reference to the rights of human subjects; the other is as a subject, regulated by law. Artificial intelligence, especially the third category of artificial intelligence, can be self-learning, capable of reasoning and decision-making, how to carry out the regulation of the law becomes a difficult problem. The current artificial intelligence is approaching the goal of “how to achieve human intelligence on computers” from one aspect or the whole aspect. Especially under the condition of accelerated development of technology, it is not impossible to realize human intelligence and recreate a kind of intelligence. The ultimate goal of artificial intelligence is to create “intelligent people”. Therefore, for this kind of artificial intelligence agents with general intelligence, how to regulate them by law, how possibly becomes the fundamental challenge of artificial intelligence for the law. This is reflected in the legal problem of whether AI can be taken as a subject of law.

This question is the most central issue of AI research. Whether AI should be taken as a legal subject becomes a prerequisite for the regulation of AI by law. The different qualifications of legal subjects, the means and ways of regulation of artificial intelligence by law also have a large difference, while all the rights and obligations of artificial intelligence are also different. This problem is also the core issue that legal practitioners should think about. And the research of artificial intelligence in a certain sense is in the creation of new social activities of the subject, these activities may have a part of human intelligence and ability, but also may have the general human intelligence and ability, and nowadays artificial intelligence also tends to have all the human intelligence this direction, that is, whether the general artificial intelligence can be recognized as “human”. Therefore, the legal characterization of artificial intelligence agents should also become an important topic of legal research.

And this challenge has reflected in whether the artificial intelligence agents can be taken as a legal subject. This problem might to be the most central legal issue in artificial intelligence research. A prerequisite for the regulation of artificial intelligence by law is to decide whether artificial intelligence should have the qualifications of the legal subject and become the prerequisite for the regulation of artificial intelligence by law. Different legal qualifications have different methods to regulate the artificial intelligence, and different rights and obligations will be possessed. Artificial intelligence research is essentially to create a new “person”. The “person” may have a part of the human knowledge and ability, may also have the general human knowledge and ability. Today artificial intelligence is also developed to be equipped with the general intelligence of human. So, can the artificial general intelligence agent be considered to a “person”? It’s a vital topic for legal scholar to research legal entity qualitative problem of artificial intelligence.

3 Artificial Intelligence and Legal Subject

The discussion of the law on the qualification of subjects has always been an unavoidable subject in all branches of law and jurisprudence. The provisions on the subject of law are mainly concentrated in the civil law and even the constitution of each country. Civil law is called “the mother of all laws”, and many legal principles and values are derived from civil law, and the provisions of civil law on subjects, legal personality and legal capacity have become the basis of all sectoral laws.

3.1 Relative Description of Legal Subject

The main subjects of law stipulated in the General Principles of Civil Law of China include: naturally born person, legal persons, and unincorporated organizations. The establishment requirements of each subject are stipulated. Naturally born person is equipped with civil rights capabilities since he was born and lasts till his death, he can exercise his legal rights and undertake his obligations. As a living organism, a naturally born person has the capacity of rights, which is the basic element to become a legal subject.

Legal person are organizations that are equipped with capabilities of rights and ability to act independently according to law which are established legally and have their own names, institutions, locations, properties and fund. Rules about the third type of legal subject in civil law are assemble to legal person in some aspects and the difference is clearly regulations about the third type is that whether qualification of a manager is needed when organization is established. The function of a manager like the department which makes policy for legal person. In this way, both these two types of organizations need to initiate an institution to make policies, perform and legally represent them to qualify legal subject. Naturally born person is the same when it comes to expressing wills. This means that only when a person or organization can clearly express themselves can they be qualified legal subject, then rights and obligations.

The criterion to judge whether a person could be legal subject in law is capacity of rights, which he is entitled to exercise rights and do corresponding obligations. Capacity of rights is a kind of qualification or ability which means legal subject is allowed to do something and forbidden to do something according to law, it can be regarded as legal personality. German Civil Code started with the chapter of personality and the first rule is that given birth means the beginning of capacity of rights. In this way, legal personality equals to capacity of rights. What is legal personality? In the System of Contemporary Roman Law by Savigny, personality is responsible person in legal relationship, and the essence of it is capability to establish a certain connection in relationship. Legal relationship is exactly the connection between personalities.

However, from the perspective of the History of Law, the concept of legal personality and capacity of rights cannot be the same. Let’s take an example of personality in Roman Law. Personality doesn’t equal to capacity of rights in Roman Law, it is the identity that support the whole system of Roman Law. A person could only be allowed to do something when he was in certain status, like father’s rights toward children and wife in family. Not every organism or naturally born person had full legal personality, besides, if someone had violated the law, the consequence might be the decrease of personality. Therefore, a person had to qualify the identity standard by law to possess legal personality. Professor Junju Ma thinks that the distinctions of term Personality is consequence of different status between different identities. The result of distinctions in law is the separation of organism between legal person. Personality in Roman Law clearly described its essence. Furthermore, the separation of a person and personality means there will be issue of determining whether a person is qualified to have personality in law. It means as follow: what kind of organism is qualified to be a person with legal personality and under what circumstance. The term personality as well as legal technique objectively reflects ancient Roman’s view of world and value. Consequently, as the essence of personality, identity legally depart a person and identity on the one hand, on the other hand, it makes personality a legal term with content of ethical value.

With the history of Western world in middle age and the Renaissance, human began to understand and realize the knowledge of themselves. Status of human had improved, and the content of personality include ration and moral values. Therefore, Karl Larenz thought that every person is equipped with capability of rights because everyone in essence is a person with full personality in ethic world. General ration and ethic require the law to admit that every person have capability of rights no matter what. Savigny hold the same opinion, he argued that since all the law was the moral and ethical being which meant that law was just the opinion for every person of the world in their inner free mind, so the original concept of legal subject or personality should be same as concept of human. Personality is more general without identity and become symbol of free will and ration. Comparing with capacity of rights, personality is more abstract which can describe legal status and qualification of subject more generally. Thence, Hegel said that personality included capacity of rights. And personality was so abstract that it was the foundation and being of essence and formality of law.

Therefore, concept of personality and capacity of rights in legal term is not exactly the same. Personality is more of ethical and moral meaning while capacity of rights is more of legal technique and less ethic. After the extract and abstraction by law, personality lost original content and only remain its elements in formality. The word Capacity of Rights was created in German Civil Code, it’s both used for naturally born person and legal person (corporation, company, etc.). Community is included in legal subject, there is no doubt that capability of rights is empirical law term. In this way, the standard to determine whether a person is legal subject is to examine whether that object qualifies capability of rights. Capability of rights, as the consequence of development of law lately, is a purely legal technique term in empirical law, it also leaves room for expanding scope of legal subject. Under this circumstance, community can be legal subject with capability of rights and the scale of subject with capability of rights will increase with abundance of social life.

3.2 The Philosophical Connotation of Legal Subject

The substantive content of legal personality would be supposed to be recovered in German classical philosophy, which has nourished the German Civil Code. In this respect, Kant’s philosophy provides sufficient nutrition for our current legal theory. It is the main content of Kant’s philosophy to think about “what is man”. And this content also enriches the connotation of legal personality. It will be better to respond to our legal status of artificial intelligence when we understand the form and substance of legal personality.

Kant argues that reason is the ability of our mind to be able to integrate empirical material into knowledge. Any knowledge consists of two parts: form and material. All knowledge begins with experience, but not originate from it. Empirical knowledge gain from our innate sensations, intuitions, perceptual consciousness. But the content which we gain is accidental. But the form or the frame structure is innate. Therefore, knowledge is integrated form the sensible material, which come from acquired experience, through the intellectual initiative to regulate and integrate. Moreover, Objective things constitute a cognitive object only through the active role of the subject. Without the participation of the subject, objective things cannot be said to be objects of knowledge. All knowledge begins with experience. Once begun, in turn, we have to reflect and analysis on such accepted knowledge, and know what it is or not. Therefore, Kant said, “Thinking without intuition is empty; intuition without concept is blind”. In the process of cognition formation, a man does not simply passively accept emotional material outsides, but active in combining the external sensible materials with the innate form of the subject itself. Through “transformation”, “provisions”, “comprehensive”, “integration” actively, knowledge can be gained. In the course of object conforming to concept, human become the legislators of nature. So that, the ability to gain knowledge is that we have the ability to specify the perceptual material through our a priori intellectuality. Kant incorporates the unity of cognition and general necessity into the subjective initiative of self-consciousness, and exclude interference from God beyond our cognitive ability. It’s really the first time to establish a person the dominant position in the nature in theory.

Thus, what is core of subject is the ability to legislate in this way, and this ability to legislate lies in the reason of the self and the ability of the self to unify sensual phenomena. The process of man legislating nature is also the process of forming knowledge by using reason. As a rational subject, man should have this legislative ability. This ability is expressed in two aspects, the innate ability of knowing and the acquired experience of perception. Knowing is unity, a kind of self-consciousness, which can unify and integrate all the materials of experience, and is a kind of subject’s ability. The subject is able to unify and integrate the acquired perceptual material through logic and other means, to know, and to acquire knowledge. Thus, in the field of cognition, man is the legislator of the natural world and knows the world objectively in conformity with the subjective through the unifying activity of knowing reason.

The ideas of classical German philosophy, represented by Kant, were digested and absorbed by the German legislation. The doctrine of the subject of Kant’s philosophy provided the theoretical basis for the determination of the legal subject, personality and capacity of rights in Germany. The free will of man becomes the ethical basis of personality, and it is due to the property of free will of personality that personality becomes the source of all laws. In the social state, reason sets down those principles which are in accordance with distributive justice and those rights which are acquired on the basis of long-term use and are guaranteed by positive law. Among other things, the spirit of subject matter is the spiritual foundation of the entire German civil law, and this spirit of subject matter is embedded in the free will and rational spirit of man and personality. Personality is also a bearer in legal relations, and free will and reason become presupposed conditions in legal relations. The formalized law also mechanicalizes those concepts that have ethically significant content, and in this sense, personhood and capacity for rights are equated. And the right capacity, as the formal meaning of personality, becomes the subject of rights and obligations, and loses the original ethical meaning of personality. Personality also becomes not only the basis of civil law, but also connected with the basic law and the constitution, and becomes synonymous with freedom and dignity, and personality takes on a sacred character.

3.3 Artificial Intelligence and Legal Personality

One can be called a “legal person” because he has capability of rights, and the substance of capability of rights derives from his personality. Personality is the ability of a rational and free willed subject to regulate and legislate. In other words, in law world, in the vision of law, personality is the ability to establish, change, eliminate legal relationship. But what kind of artificial intelligence can be capable of this? According to Kant’s theory, it provides us with a beautiful blueprint for artificial intelligence. Kant divides the process of human cognition into two parts, that is, an integrated analysis of sensible experience and rational faculty. This rational faculty to unify perceptual experience, the most significant way of which is what Kant calls a priori logic. Logic is regarded as regular pattern and approach to learn knowledge and the most important methods to realize intelligent in the aspect of artificial intelligence. The conclusion that can be made without any exaggeration is that Kant’s philosophy has made full preparation for artificial intelligence at this point.

As for artificial intelligence, this kind of system might include two parts. One is the input of external sensory material. The other is the analysis of the input of the inductive material. That is perceptual ability section and intellectual ability section. According to Kant ‘s theory, the intellectual ability section could be able to reflect the ability of artificial intelligence as legal personality. Its main task is to organize these input materials into a coherent, meaningful world experience. And intellectual ability is to find out the law of human cognition. And Kant think, logic is the law of thinking, so he created the transcendental logic. And the study of logic is further in-depth. And the logic applied for the artificial intelligence might be different from our human. Because it’s essentially different between man and artificial intelligence. But the rules of thinking might be common, and the artificial intelligence is designed to imitate our mind. So, in that case, the Kantian theory provides great blueprint and fine cognitive model. Our cognitive process generally contains two elements. One is low-level cognition, which is information processing from different senses. The second is high-level cognition, which is that the subject, through the access of concept and idea, understand the situation in the conceptual level, extract the meaning from situational experience. With object recognized into the abstract formal framework, situation would be understood as a whole.

In addition, there is huge difference between legal person and naturally born person. The subject of legal is a hypothesis based on rational people. A rational person is formalized through the law, excluded from personality, preferences, emotions and other irrational factors. Because the legal positivism made the study of law more professional and become a strict subject. Such rational person is more adaptable for artificial intelligence. Therefore, from this level, more artificial intelligence can be given the status of the legal subject.

In this case, it is important to note that the legal subject of artificial intelligence is philosophical abstract status. That is, artificial intelligence has the possibility of becoming a legal subject. This legal subject status is also abstract, is the possibility of becoming a legal subject. Artificial intelligence agents to meet the conditions discussed above, it is possible to become the subject of the law and as for what kind of subject, it’s required in the specific legislation to be clearly defined. In other words, it’s prerequisites and prerequisites for artificial intelligence becoming legal subject to have the rational ability for natural legislation and the ability to regulate self-control. And this abstract legal subject on the one hand is reflected the influence on our current legal system, and out daily life, on the other hand is a response of law on intelligent agent with similar capabilities. Therefore, the artificial intelligence agent can become the legal subject is abstract, only to express the possibility of artificial intelligence agents have become the subject of law. This subject may be a fake “person”, a legal person possibly, or another subject possibly. Similar to the “electronic people” in EU legislation, the specific legal status of artificial intelligence needs to be responded and clarified in specific legislation.

4 Non-axiomatic Reasoning System and Legal Personality

Non-Axiomatic Reasoning System, short from NARS, is designed and created by Pei Wang, a professor of Temple University. NARS is a computer reasoning system for achieving general intelligence. And NARS can learn from interaction and experience. And faced with unspecified situation and questions, it can do operations and answer different questions.

4.1 NARS and Its Basic Reasoning Rules

The design of NARS is based on the belief that the essence of intelligence is the capability to adapt to the environment and to work with insufficient knowledge and resources. NARS uses a new form of term logic, or an extended syllogism, in which several types of uncertainties can be represented and processed, and in which deduction, induction, abduction, and revision are carried out in a unified format. The Assumption of Insufficient Knowledge and Resources has become the logical premise of the NARS, and under this assumption and design, NARS has the following characteristics.

Finite: the system’s computing power, as well as its working and storage space, is limited;

Real-time: the tasks that the system has to process, including the assimilation of new knowledge and the making of decisions, can emerge at any time, and all have deadlines attached with them;

Open: no restriction is imposed on the relationship between old knowledge and new knowledge, as long as they are representable in the system’s interface language.

Adaptable: the system can accommodate itself to new knowledge, and adjust its memory structure and mechanism to improve its time and space efficiency, under the assumption that future situations will be similar to past situations.

These characteristics are the basic features of the NARS and are different from the current mainstream artificial intelligence. In order to achieve the goal of artificial general intelligence, to achieve rational, NARS reasoning system has the following components:

A formal language for the communication between the system and the environment;

An interpretation of the formal language that makes its sentences correspond (maybe loosely) to human knowledge represented in natural language;

An inference engine with some inference rules to carry out tasks, such as match questions with knowledge, generate conclusions from promises, and derive subquestions from questions;

A memory that stores the tasks to be processed, and the knowledge according to which the tasks are processed; and

A control mechanism that is responsible for the choosing of premise(s) and inference rule(s) in each step of inference, and the maintaining of the memory.

NARS is composed of these basic parts, the core of which lies in the first three parts. The first three parts constitute the basic reasoning system of NARS. NARS reasoning system is to rely on the logic, which is to study the rules of thinking, and apply the method of logic to achieve intelligence. NARS logic is different from our daily formal logic, is a non-axiomatic logic (NAL). NARS has nine logical reasoning system configurations; each level has its own syntax and inference rules. And the reason why the whole system is not axiomatic, is that the designer of the system, don’t fix the knowledge base of the entire system, and don’t give the knowledge base a proposition in the form of axioms, after preset after each level of reasoning rules of system. And all the knowledge of NARS can also be improved and enriched constantly based on the experience and interaction of system. NARS also constantly interact with the outside world in its open structure and obtain new knowledge and awareness under the guidance of the initial reasoning rules.

The NARS has modified and expanded the traditional classical logic, using a new set of non-axiomatic logic under the assumption that knowledge and resources are relatively insufficient. NARS development takes a gradual approach, at each stage, the logic is expanded, giving the system a more expressive language, richer semantics and greater reasoning rules. In general, NARS has the four stages of NAL, including basic reasoning, first-order reasoning, high-level reasoning and procedural reasoning. Among them, the most characteristic of NARS is the way of identifying things. That is, the truth-value function of NARS. For a given statement, if the amounts of positive and total evidence are written as w + and w, respectively, then the truth-value of the statement is represented as a pair of real numbers in [0, 1], < frequency, confidence >, where frequency = w + / w, and confidence = w/(w + k). (k is a constant, it means that NARS compares its current evidence to the future evidence. For example, if k = 1, NARS always compares this evidence to the next time.) This is not only the way of expression, but also a judgement of object and things. The judgement will be a kind of knowledge. Thus, NARS interprets a judgment in w instances, where the positive instances have w +, and the confidence is w/(w + k), and NARS believes the statement at the extent of w/(w + k).

Consequently, truth-value in NARS uniformly represents several types of uncertainty, including randomness, fuzziness, and ignorance. Defined in this way, Non-Axiomatic Logic has an experience-grounded semantics where the meaning of each term and the truth-value of each statement are determined according to the relevant experience of the system, by the system itself. So, the way of cognition of NARS is similar with our human.

The truth-value function of NARS is common in structure of cognition and judgement. This innate structure of cognition is relevant to logic, but the content of cognition and knowledge is connective with experience. Only in the participation of the material of experience, a judgement with content can be constituted. NARS contains other reasoning rules in addition to the basic reasoning rule of the truth-value function.

The basic reasoning is mainly through the truth function to form knowledge and judgment. First-order reasoning is to extend the non-axiomatic logic in the way of adding the knowledge and judgment obtained in the previous stage into the rules of language and grammar; through the way of set theory to define the relationship between different judgments. High-level reasoning is to allow statements to be used as terms, through deduction, induction, abduction, syllogism or other aspects of the way to high-level reasoning. In addition to the above three levels of logical reasoning process, another special feature of NARS is procedural reasoning, the fourth stage of reasoning. In this stage, time has been introduced. Procedural reasoning means to infer about events, operations, and goals. At this stage, an event is defined as a special type of statement that with temporal truth-value, that is, its truth-value can change over time. Operations are the means for the system to achieve goals. NARS can carry out some external operations and internal control. The execution of an operation is usually accomplished by a mechanism outside the reasoning system, such as a hardware device with certain sensor/effecter capability, or another computer program with certain information-processing capability. The reasoning system interacts with them by issuing execution commands and collecting execution consequences, and these activities consist of the sensorimotor processes of the system as a whole. Operations toward the outside of the system correspond to perception and control of the environment. Operations toward the inside of the system correspond to self-perception and self-control. Therefore, NARS is not only to absorb new knowledge and answer new questions, but also to achieve a variety of operations, in the process of interaction with the external. Under the original goals and independent goals, NARS can uniformly handle goals/operations and ordinary tasks/beliefs by modifying the internal architecture and control mechanisms, and ultimately can become a relatively independent artificial intelligence reasoning system.

4.2 NARS and Legal Personality

Artificial Intelligence system is totally different from a person in essence. Human is natural creature with many special biological features. Our brain also evolves to adapt the change of nature. But artificial intelligence system does not develop as human does, it is the simulation of human’s brain which in essence is a digital system without any biological features. Artificial intelligence system like NARS do have something in common with human. Can NARS be a legal subject?

The key of status of being legal subject is rational capability and freedom to legislate for the nature and human being themselves. Human as subject is foundation and origination of all regulations, which means that subject is capable to set regulations and orders, and this kind of capability is absolutely initiative and beyond themselves to legislate. This capability is basic content of personality, and it is the necessary element of qualifications to be legal subject.

The mode of cognition of NARS is based on a set of truth-value function which NARS can understand and make judgment with available knowledge. Truth value of NARS is defined with F (frequency) and C (confidence), and both these two values are gradually formed and modified in the process of interacting with externals. Therefore, NARS is directly related to experiences.External experiences are input into NARS and the mode to deal with data is NAL. These reasoning rules have many levels and ranks to deal with different data separately. There is basic logic to understand and judge the certain subject, and elementary reasoning rules to handle the relationship between subjects, and high order rules for different sentences and different variables.Reasoning rules which related to logic and set theory and math in themselves are predesigned in NARS and gradually become the method for NARS to know the external world. In this way, truth-value function of NARS is composite of innate logical reasoning rules and acquired experiences. The content of subject and knowledge comes from experience in the design of innate logical rules. This kind of cognitive mode happen to be same as Kant’s intellectuality. With cognition and integration by NARS, knowledge about materials from externally objective experience can be shaped and modified finally. Therefore, NARS is capable of cognitively understanding this world and own knowledge at this point.

Besides, NARS system is capable of regulate itself and achieve its goal. In NARS’s reasoning process, the system itself can make procedural reason, make judgment of events and operations and goals, which means that the system does not only learning to observe, it also operates. a goal is defined as a statement to be made true (or as close to true as possible) by executing operations. In this way, a goal can be represented as an event, whose preconditions and postconditions can be gradually revealed through reasoning on available knowledge, and eventually related to operations. For certain operation, the system can describe the preconditions (causes) and postconditions (consequences) of a given operation, and carry out explaining, predicting, planning, and skill learning by reasoning on these descriptions.The original goal of NARS system can probably come from maker’s design, or can be entitled by its users or managers which is pretty random for the system. But once original goal is given and set, designer cannot have control over other derived goals by system itself. The realization of every goal need integration and qualification of many elements, NARS system can illustrate the condition and process of realization of goals by reasoning rules. All available conditions and elements can be new goals of NARS system. With these reasoning process carrying on, original goals are not important anymore because new goals can independently be system’s goal. In this kind of process, NARS system can absolutely control itself. In the meantime, when the goal of NARS system was defined, all its operations require acquired learning. The system can only know procedures to operate through experiences. After many times of operations, NARS has to acquire skills to operate, and more importantly, it has thoughts and can reflect itself. In this way, NARS system will gradually know that it is different from external world and then it will realize independence itself by reflection. The sense of self for NARS will finally shape in this process. The reasoning system exerts control by issuing orders and collecting results of former orders. Operations for external world correspond to knowledge and control of environment, and operations for internal system correspond to self-cognition and self-control.

The former several levels of reasoning rules make preparations for procedural reasoning and form knowledge base for system itself. The system establishes and carries out goals and plans in this knowledge base by decision-making procedure and realize its own goals. In this process, reasoning system keeps collecting and issuing orders and achieves self-control and management. The operation is the reflection by system itself and NARS also needs this kind of reflecting-operation to avoid making mistakes and even assess afterwards. It is the reflection that makes NARS realizes its existence by itself. Furthermore, reasoning system also controls external devices’ operations to contribute to realization of goals. This means that the system can sense and control external condition, and in the process of goals’ realization and expectations, knowledge and control about system itself become deeper. It makes the sense of self and self-regulation of system more prominent through dividing the insides form the outsides and the subject from the object. So that, the system has achieved the self, but also regulated the self, also have the ability to create, change and eliminate a specific social relationship or legal relationship.

In result, NARS has such ability of awareness and execution, which is the essential connotation of legal personality. Especially, legal personality is abstracted and formalized. Through legal positivism, and gets rid of ethics and morals gradually. It’s necessary condition for NARS to becomes a true legal subject. Therefore, NARS can be given the qualifications of the corresponding legal subjects, with the capacity of rights.

5 Conclusion

Law and jurisprudence must respond in a timely manner to the profound impact of artificial intelligence on society, particularly general artificial intelligence, and the many implications and risks that go hand in hand with such challenges. And with many social sciences mired in division and controversy, only law can respond institutionally on behalf of humanity. Among the many tasks that need to be accomplished, one of the first steps in the legal regulation of AI is the legal positioning of AI. Whether artificial intelligence can be taken as a legal subject, whether it enjoys the subject qualification of assuming rights and performing obligations. The legal subject or the capacity of rights is closely related to the rational capacity of human beings, so it becomes necessary to explore the conditions and the possibility of the qualification of AI to be taken as the subject of rights from the level of legal philosophy in the abstract. With the rapid development of technology and the in-depth research of artificial intelligence, there will be more and more such artificial intelligence with rational ability, and artificial intelligence will gradually become an important part of our society. The law should also take more comprehensive consideration of the characteristics and capabilities of AI in order to better play the role of law. In the future, in an era of human-machine coexistence, the law will not only be a law for people, but also a norm for other intelligences.