Preface

Intelligence is currently the hottest word in our society. However, experts and scholars in related fields have no agreed-upon definition for it. Although most of us believe that our society is moving toward the era of intelligence, we are not sure what can be called intelligence, where its starting point is, and what patterns its occurrence and development follow. These questions appear to become a prohibited area for research. More than one book on intelligence or artificial intelligence (AI) warn us that no principles exist for intelligence. When I tell my friends in this field that I am going to write a book on principles of intelligence, I can see doubts and concerns in their eyes. This makes me to be more cautious in doing research on principles of intelligence and for which I should thank those friends of mine. No matter how bumpy the road to exploring principles of intelligence might be, I shall try my best to tackle this problem as if there were an invisible hand pushing me and a voice urging me. Even if my research only lays one pebble for this field of study, my efforts to it would not abate.

In over past 30 years, my pursuance of research in intelligence has been intermittent but has never been abandoned. In 1984, I completed my master's thesis titled "Rule-based automatic indexing of Chinese documents in science and technology", which was essentially an expert system. At that time, however, I knew very little about expert systems. Based on this essay, Applied for funding for a doctoral research project from the Ministry of Education. As the principal investigator, I worked with six lower class graduate students. This system received unanimously positive feedback. It is this experience that cultivates my interest in artificial intelligence and expert systems.

After graduation, the nature of my work did not allow me to focus on research. But I did not stop learning and thinking about the problem of intelligence. In 1990, I had time to edit a book titled "Expert systems and their applications in management” which was published by Tsinghua University Press in 1991. Since then, my personal interest in intelligence has not diminished. Instead I would read related documents, take notes, and analyze research results and approaches of different disciplines in order to determine the direction of intelligence research. In May 2009, I gave a talk entitled “From mid-18th century to mid-22nd century:Evolution of human civilization" at Sun Yat-sen University in Guangzhou. I predict in the talk that the time in which people work for wages will not exceed 10 hours per week. Robots will be the main workforce for material goods production, shaping the modes of industrial and commercial production and daily life services. All countries will have basically completed the reconstruction of such an industrial system. "I also predicted that" with the development of intelligent technology, biotechnology and information infrastructure, humans and non-biological autonomy will coexist in various economic and social activities. The role of the latter will continue to grow with the deepening of information civilization and establish a structure of societal governance in the environment of multiple intelligent autonomy. This constitutes an important task for the transition from industrial civilization to information civilization as well as the touchstone for testing the rationality and wisdom of mankind." In 2011, I delivered a speech titled"Trilogy of information revolution: Emancipating human mind, extending human intelligence, and intelligence independent of humans" at Baidu. In October 2015, I specifically talked about at the China Institute of Information Communications why the next stage of societal development was the era of intelligence. I have also been involving in the planning and implementation of some smart cities and intelligent manufacturing projects since 2009. It is my research, observation,thinking and practice in over 30 years that forms the foundation of this book.

We now can see AI systems carrying out conversations, competing on the same stage, playing chess, and accompanying humans to walk. Robots are in charge of many jobs in production and service areas such as machinery processing, graphic design, food making and serving, stock investment, patient diagnosis, language translation, and speech recognition. The same or similar results or behaviors are from different subjects (i.e., humans or non-biological autonomy) using different modes of implementation. What relationships lie behind such phenomena? Whether a unified theory of intelligence can be established? These are the goals I have been pursuing for many years. Now that I am retired, I finally have the time to immerse myself in this research in the past several years and consequently have completed my book on the topic.

The book consists of seven chapters. Chapter 1 is a review of research in fields related to intelligence, aiming to derive a definition of intelligence and identify major factors that determine its occurrence and development. Chapter 2, built on the basis of Chapter 1, defines intelligence, describes its major components before setting a framework for further discussions. The framework serves as a general outline for the theory of intelligence. Chapter 3 and Chapter 4 explain the evolution, development and use of intelligence based on the definition and framework provided in Chapter 2. It also attempts to draw conclusions that are of regularity. The analyses performed in these two chapters show that the theory of intelligence I presented in Chapter 2 can fully explain the process of intelligence evolution and development as well as the application and practice of various types of intelligence up to present. Chapter 5 and Chapter 6 are further abstraction and generalization of the main conclusions drawn in Chapter 3 and Chapter 4. Specifically, Chapter 5 summarizes and analyzes the 10 logical features of intelligence, from which 10 norms of intelligence are induced. Chapter 6 proposes a semantics-processing-based computation framework for intelligence on the basis of Chapter 5. According to the conclusions of the previous six chapters, Chapter 7 conducts schematic analyses of how human society pushes forward the evolution of intelligence into its final stage. This chapter also depicts characteristics and key paths of intelligence and its developmental rationality that concerns us. Chapters 5-7 are written on the basis of the theory of intelligence proffered in Chapter 2, which illustrates the theoretical framework's predictability.

Chapter 1 of this book provides a comprehensive and systematic description of the major achievements of research on intelligence in related fields, which serves as the foundation for the entire book. This chapter reviews the major contributions of these fields from four different perspectives. First, it is about the enlightenment brought to us by the pioneers of human civilization in their understanding of spirit, soul and wisdom in the past thousands of years. Second, from the perspective of evolution and development of biological intelligence, this chapter presents research findings on biological intelligence from the fields of evolutionary biology, basic life science, molecular biology, biochemistry, plant physiology, animal behavior science, cognitive neuroscience, neuropsychology, cognitive psychology, developmental psychology and more. Third, from the angle of non-biological intelligence, I discuss the contributions made to intelligence research from simple tools, non-digital machinery, computation tools, digital devices, automation systems, artificial intelligence, logic, computation and other domains respectively. Fourth, from the perspective of cross-disciplinary study of biological and non-biological intelligence, I analyze the research results represented by mind research. Through the analyses and discussion from the above four perspectives, some important patterns of the evolution and development of biological intelligence are observed. The commonality shared between biological and non-biological intelligence is also discerned. The findings from all those disciplines and domains together provide a foundation for the definition and basic framework of intelligence posed in Chapter 2.

Chapter 2 is the core of the present book. It is based on summarizing all the research results of related fields that this chapter puts forward the definition,components and interrelations of intelligence. Chapter 2 functions as a general outline for this book. Intelligence is such a real, objective existence. It has such a close relationship with human survival and development. In addition, research on intelligence spans thousands of years and there are countless research documents about intelligence. Yet the debate on what intelligence is never ends. One school insists that the intelligence is psychological or spiritual. However further research inevitably finds material characteristics in the development of intelligence, and such characteristics cannot be explained by spiritual mind. Another group insists that intelligence is material. Nevertheless research on the topic would eventually run into psychological or spiritual features of intelligence that cannot be explained by today's neuroscience and brain science. The study of mind attempts to go beyond both the psychological/spiritual view and material view. But the explicability and predictability of the third view is not good enough for all to accept.

The definition for intelligence in this book is based on the one in The MIT Encyclopedia of the Cognitive Sciences (MITECS) with slight adaptation:"Intelligence is the subject's abilities to adapt to, change and select the environment." The subject here includes both organisms and non-organisms, which is consistent with the MITECS' definition. From this definition, I further define the major components of intelligence: autonomy, function, information and environment. The environment is an important external factor that affects the evolution, development and use of intelligence. I also define what constitutes entity, function and information. The definition, major components and composition of intelligence together form the basic framework of intelligence theory.

Chapter 3 determines the starting point of intelligence, analyzes the first five stages of intelligence evolution, and examines the development process and characteristics of biological and non-biological intelligence. According to the definition provided early, this book sets the starting point for intelligence to be the first primitive life entity on earth. No matter how simple the intelligent entity is, it is capable of adapting to the environment and remaining in existence as an entity. This chapter analyzes the evolutionary and developmental process of the main types of biological and non-biological intelligence through the main line of entity, function, information and major environmental influence, identifying the six stages of intelligence evolution: single cell organisms, nerve systems and brains, languages and writings, computation and digital devices, automation and intelligence systems, and non-biological intelligent autonomy.

The evolution and development of intelligence display two main lines, one is biological intelligence, and the other is non-biological intelligence. The evolutionary path of both has the property of moving forward toward opposite direction. Biological intelligence develops from very low function and information capabilities on the basis of autonomous entities while non-biological intelligence develops toward autonomous non-intelligent entities through the human-given control ability on the basis of practical functions and information processing capabilities. Both begin crossing in automation systems and artificial intelligence systems and overlap at the last stage of intelligence evolution -- non-biological intelligent autonomy. This chapter uses the major components in the intelligence composition framework Chapter 2 posed to perform a comprehensive analysis of the first five stages of intelligence evolution as well as the two main lines. The intelligence theory I proposed indeed is of good explanatory capability. The evolution and development of intelligence in essence all advance to a higher level of intelligence. The purpose of differentiating the two is for defining the two advancing modes with different characteristics in order to better understand the pattern of intelligence development. Development refers to the intelligence increase in one life cycle of an intelligent entity. In contrast, evolution means intelligence development spanning the life cycle and individual intelligent autonomy. In one life cycle, biological intelligent autonomy go through the learning, use and transmutation processes while non-biological intelligent autonomy undergo the processes of given, use and elimination. These major changes also occur in entity, function and information.

The focus of Chapter 4 changes from the forming and components of intelligence in the last two chapters to how to use it. First of all, this chapter performs a panoramic analysis of the objects of intelligence use: intelligent events and intelligent tasks. Intelligent events and intelligent tasks are different names of the same concept in different settings. Intelligent events refer to all kinds of matters that actually exist in intelligent autonomy in all areas of the society. By comparison, intelligent tasks are various tasks to be executed by individual intelligent autonomy. Executing intelligent tasks can be regarded as solving problems. Intelligent events or tasks with different features have roughly the same problem solving process. But there are different strategies and paths for problem solving. For any intelligent entity, the completion of problem solving is not to obtain an answer or outcome. What is more important is the learning after the evaluation of the problem solving, the development in understanding the process and outcome. Therefore, the process of intelligent use is also the process of intelligence development.

Chapter 4 also proposes the main indicators for evaluating intelligence from three perspectives. One is the complexity of intelligent events or tasks. The other is the readiness, maturity and completeness of intelligent autonomy. The third is the macro-effect, effectiveness and growth of intelligent use. By systematically analyzing the types of intelligent tasks, strategies and paths of problem solving, this book reaches an essential conclusion: when solving the same or similar problems, the more algorithms and computations the intelligent entity uses, the less intelligence maturity the entity has.

What is discussed in the first few chapters includes all intelligent autonomy while in Chapter 5 and Chapter 6 non-biological autonomy become the subject of examination. Chapter 5 describes 10 logical features or norms of intelligence. Intelligence is semantic in that intelligent autonomy possess information, and information processing is based on semantics rather than symbols that carry semantics. This is the main reason that leads to all intelligence logic and computation features. The intelligence that an intelligent entity possesses is comprised of individual, specific components of intelligence. Components and connections are the direct manifestation of semantics, which constitutes the main form of all semantics-based intelligence processing. The three norms of overlay, decrement and cross-subject integration form the basic computation methods for the evolution and development of intelligent components. Fault tolerance retains diversity and possibility, and specifications guide all intelligent components as a whole to become rational, making the right provisions for intelligence rationality.

Chapter 6 posits a semantics-processing-based intelligence computation framework. Unlike the symbolic-processing-based computation framework by John von Neumann, intelligence computation is based on semantic logic and aims at the sustained growth of non-biological autonomy' intelligence via the process for the computation of internal and external intelligent tasks. Its operation is triggered by external aesthesis and internal computation, goes through the cycle of strategy determination, resource reallocation, task execution, process evaluation, achievement learning, and intelligence expansion, and finally forms an intelligence computation cycle based on the process of intelligence behavior. On the basis of this process, non-biological autonomy' intelligence gradually increases. Intelligence computation framework consists of three parts, one is the construction of intelligent behavioral processes which includes triggers and distributors, strategy generators, actuators, and evaluators. The second part covers resources of intelligent autonomy that involve intelligence components, microprocessors, computation resources, and behavioral resources. The third part of the framework encompasses the environment of intelligent autonomy that contain external events and resources. After the intelligence computation framework is presented, Chapter 6 specifically makes some in-depth analysis of its main components, micro-functional units, functional unit groups, functional systems and beginning point and growth process. Description components are the carriers for all functions and information of non-biological autonomy, replicated genes, and memories that can be called by intelligent behaviors. Connections and description components together constitute information semantics. External aesthesis, connections, description components, micro-processing and computation form the mechanism for intelligence computation framework to continue moving toward perfection. Different from the symbol-based von Neumann framework, the intelligence computation framework has unique and distinctive computation architectures, especially the micro-processing and internal computation. This is the most characteristic part of the intelligence computation framework. The intelligence computation framework performs nonstop computation before reaching its perfection. It grows through computation. The internal computation will not stop until exhausting all internal paths and finding no new learning materials within or from outside.

Chapter 7 illustrates the paths regarding how the evolution of intelligence enters the sixth phase -- non-biological intelligent autonomy - and the prospect of its entering into the sixth phase along with some analysis. There are three major changes when non-biological intelligent organisms evolve from the current automation systems or artificial intelligence systems to non-biological intelligent autonomy: (1) autonomous control intelligent behaviors, (2) self-learning and growth, and (3) autonomous acquirement of computation and physical resources. This chapter analyzes the possibilities for intelligence to reach the sixth stage. It also discusses the goals, paths and key tasks respectively in terms of entity, function and information - the three major components of intelligence. Particular emphasis is placed on control functions, logic of learning modules, and genetics of non-biological intelligent autonomy, that is, construction of complete function and manufacture tools, production lines or other infrastructure of information components for intelligent society development. In addition, Chapter 7 in particular considers how to govern the society composed of humans and non-biological intelligent autonomy, and how to identify and realize the rationality of non-biological intelligent autonomy in order to establish social norms and guidelines that both humans and non-biological intelligent autonomy would observe. Meanwhile, it must be explicitly pointed out that there is no need for humans to be too concerned about the likelihood that non-biological intelligent autonomy would surpass human intelligence because the essential point for judging this new society is that intelligence growth is in direct proportion to rationality growth.

This book summarizes the major rules followed by intelligence evolution, development and use. Those rules are, on one hand, in agreement with and, on the other hand, contradictory to the well-developed and enormously successful mathematical and physics rules. For example, either humans or walking robots must follow the physics rules such as overcoming gravity and paying attention to slopes as well as wind when walking. However, why walking, why choosing this route, why walking at different speeds, and how to control the walking process? Those questions do not fall within the laws of physics as such laws cannot explain those behaviors theoretically. Perhaps mathematical logic can be applied for explaining why choosing this route, why at different speeds, why walking, and how to walk. But still only some points are explicable while others are not. Taking robot and human waiters in restaurants as an example, why robot waiters walk depends on customer needs. How robot waiters walk is based on complex aesthesis, strategies, algorithms as well as control of force and physical components. Why a particular route is taken is determined by the built-in route optimization algorithm. Why different speeds are chosen relies on the built-in route conditions and the control module of actual settings. As for human waiters, the reasons for why walking, why taking this route, and why at different speeds are the same. But the logic for completing this processes are entirely different. Experienced human waiters' decision-making for these three questions almost involves no thinking and reasoning, let alone determine how high each step should be, how to coordinate one's hands, how to keep balance. Yet all those problems must be addressed via computation algorithms for robot waiters. If we analyze this process of human waiters for several seconds, its internal cognitive processes and behaviors are extremely complicated. If we adopt the simulation approach to complete this task, more computational resources are needed than using the method based on algorithms and logic. Although deciding how to reach the table where customers sit by glancing is practically an intuitive process for the human waiter, it involves several hundred millions of synapses, each hand and foot movement would instantaneously connect hundreds of thousands of synapses between motor neurons and muscles. The reason why it is very easy for human waiters to go through such a complex process smoothly is because this entire logic process does not include any thinking and reasoning. Rather they simply complete it by intuition and experience. Intelligence maturity is not determined by algorithms and logic. Further, the intuition or instincts humans have in their evolution and development likewise are not acquired via logic or algorithms. We need to return to the fundamentals which refer to both the applicable as well as non-applicable, special relationships among intelligence, information and physics rules, and computation logic. Information representation and the essential development of intelligence do not follow physics rules but the physical components of information and intelligence observe those rules. Algorithms and logic can be used for explaining quite well symbolic processing or formal behaviors and processes of intelligence. However, existing logic and algorithms often are not appropriate for solving problems relating to semantics and autonomy. Mathematics, philosophy, and logic all inform us that any scientific conclusion applies to a specific problem domain only. Problems of different nature cannot be addressed using a universal solution. Perhaps matter can go back to the past through a space-time tunnel while the same does not hold true for living autonomy, information (where information is not what is defined in physics and instead based on the meaning of setting and connection) and intelligence (where intelligence is not a carrier of intelligence and instead is the function that relies on the carrier for existence). We should not use mathematical formula to denote the emergence and development of a major discipline. It is not because the discipline is immature. Rather it is due to the immaturity of mathematics and because mathematics does not find an approach to representing this discipline properly.

Researching on the occurrence and development pattern of information and intelligence requires both the full understanding of the pattern regarding the role mathematics, physics and life science play in information motion and intelligence behaviors as well as comprehension of where those disciplines are not applicable and why so. This is the approach to promoting and ensuring that our understanding and practice of information and intelligence are on the scientific course after knowing the essence of the problem.

My intention to study information is for research on intelligence as information and intelligence are inseparable. Chapters 3-5 in The Nature of Information are actually written in preparation for explaining intelligence. The formation and improvement of information structure is an intelligence process. Without intelligence, information cannot move toward an independent, self-existent space. Nor can it gradually complete the comprehensive description of objective objects. Without a comprehensive description of various kinds of objects, intelligence behavior based on semantics and logic would lose its foundation. Years ago I planned to write a book on "Principles of Information and Intelligence". But when I started writing The Nature of Information in 2015, I decided to split the originally conceived book into two. The main reason behind this decision is that information is matter while intelligence is thing although both are unique and objective existence. The two show great differences in both logical structure and description method.

I feel uneasy when this book is to be published. I wonder if the book has touched on the principles of intelligence and if my discussion about the occurrence and development of intelligence is in line with the reality. Such questions are coming back to me even though I have considered them time and again in the past. As a basic theoretical exploration, there are indeed many areas that need further analysis, verification and elaboration. In spite of all these, I still have this book published under the title of Principles of Intelligence, because the era of intelligence is rapidly approaching and the lack of a basic theory of intelligence is already becoming quite obvious.

This book presents a brand new logical, conceptual and description framework for the theory of intelligence. It is not hard to see the difficulty in translating its Preface and Table of Contents. I thus wish to express my thanks to Professor Heting Chu from Long Island University in the United States for providing the translation.

I also would like to thank LIU Jiuru, QIN Xujun and DONG Yafeng of the Publishing House of Electronics Industry for their dedicated work in getting this book published. This book would not have been published at such a speed and in such a fine form without their great efforts.

Yang Xueshan

2018.1