The Problem Of The Self In The Study Of Artificial Intelligence

The problem of self is important in the field of artificial intelligence research.General artificial intelligence and strong artificial intelligence is on behalf of the future development direction of two artificial intelligence research field.For general artificial intelligence,its core appeal is to pursue targets that even for its designers are unexpected and to achieve qualitatively different kinds of new task,they both requires system must have self-supervised learning and the ability of self-control.General artificial intelligence should also be able to truly interact with the world,which requires it to be aware of its relationship to the environment,and building a model of the self is crucial to this.The research on artificial emotions is also an important issue in the field of general artificial intelligence,because it not only has the function of optimizing computing capacity,but also makes artificial intelligence more like human reasoning.The subject of these emotional experiences is the self of artificial intelligence,which is obviously indispensable for the research on artificial intelligence.For the research of strong artificial intelligence,its fundamental pursuit is intentionality,and the generation of intentionality is also closely related to self.In a word,the problem of the self is an unavoidable problem in the future research of artificial intelligence,and the study of the problem of self has attracted the interest of Chinese and foreign artificial intelligence researchers.By referring to the discussion of the problem of self in the philosophical field,the main work of this paper is as follows:(1)The importance of the problem of self to artificial intelligence research is expounded.In view of the complicated definitions and schools of artificial intelligence,this paper chooses general artificial intelligence and strong artificial intelligence as the objects of study,because they are generally recognized to represent the future development direction of artificial intelligence,and analyzes the relationship between their core demands and the study of the self.(2)After demonstrating the importance of the problem of self to the study of artificial intelligence,this paper analyzes the minimal self theory in contemporary phenomenological philosophy.On the basis of reconstructing Dan Zahavi's argument that phenomenal consciousness must contain the minimal self,I then search for such a minimal self's structure in Husserl's text,that is,the structuring structure of internal time consciousness.After that,this paper analyzes Dan Zahavi's criticism of the higherorder theory of consciousness,and compares his theory of minimal self with the proposal that the self is the ability to generate consciousness and the theory that the self is the narrative construction,and points out the superiority of the theory of minimal self.(3)This paper describes the technical details of a promising non-axiomatic reasoning system(NARS)related to the problem of self,and discusses whether it has achieved the minimal self's structure and capability.First of all,this paper expounds the theoretical presupposition of NARS,one of the most important is the insufficient knowledge and resources that,in turn,combing its semantic system based on experience,using predicate logic inference rules and dynamic allocation of computing resources control system,and discusses the subjectivity of its experience and its ability to have a self.Through the above work,this paper draws the following viewpoints:(1)For general artificial intelligence and strong artificial intelligence,the study of self is of great significance for the advancement of these two fields.(2)For the self as the ability to generate consciousness and the self as the narrative construction,Dan Zahavi's theory of minimal self is the basis of the realization of former both.The minimal self also has an observable structure,that is,the structuring structure of internal time consciousness.(3)Based on Dan Zahavi's theory of minimal self,this paper proves that the inference control mechanism of NARS system contains minimum-self structure.According to Dan Zahavi's theory,this means that NARS have the potential to generate more complex selves and even phenomenal awareness.Finally,this paper also analyzes the unique self-concept acquisition method of NARS,which makes it capable of conforming to Peacock's non-conceptual first-person content holding rules,and has the cognitive ability required to produce non-conceptual first-person content.