Search

Artificial intelligence started with exploring concept of representation of real world in a database of an artificial system such as robot or expert system. The metaphor for artificial intelligence was a computer program. Many believed that Al-related computer programs should be written in specific programming language such as LISP or Prolog. That view changed with approach that representations and real world should have reality check mechanism that will interface the real world. The view shifted toward need of a body that will carry out sensors for a real time interaction. So-called embodied AI introduces that agents rather than programs should be metaphor for artificial intelligence. What is important is the agent-environment interaction, an intelligent behavior that will prove the representations. In this presentation we argue that it is time that we should pursue beyond the concept of AI. It is known in biology that limbic system includes both memory and emotions. So the concept such as motivations, emotions, and curiosity, among others should be included in agents. In short metaphors for AI should be personalities rather than agents. Not just behavioral environment, the genetic environment should be included in our theories of intelligent beings. In this keynote we will present a coherent theory of consequence driven systems toward the engineered psychology part of IT revolution.

Reading Heidegger’s Being and Time and Merleau-Ponty’s Phenomenology of Perception suggested that Symbolic AI with its representations of meaningless facts about the world could not solve the frame problem, and that the best representation of the world is the world itself. Now GOFAI has failed, and Rondey Brooks boasts that his animats avoid the frame problem precisely by directly relating to the world. But Brook’s animates and all other versions of what some call Heideggerian AI have their own version of the frame problem, viz. that the program can’t update relevance. Fortunately, there is at least one model of how the brain could provide the causal basis of such an ability. Walter Freeman, a founding figure in neurodynamics and one of the first to take seriously the idea of the brain as a nonlinear dynamical system, has worked out an account of how the brain of an active animal can directly pick up and update what counts as significant in its world. But, to program Heideggerian AI, we would not only need a model of brain functioning such as Freeman’s; we would also need a model of our particular way of being embedded and embodied such that what we experience is significant for us in the particular way that it is. This shows the task of a Heideggerian AI to be overwhelmingly difficult and casts doubt on whether we will ever be able to accomplish it.

In this chapter we argue that metaphorical utterances often contain source-domain elements that are important for the overall meaning of the utterance but that should not themselves be given a parallel in (be mapped into) the target domain, either because it is difficult to do so or because doing so is not important for linking the utterance into the context supplied by the discourse. Non-parallelism goes hand-in-hand with metaphor understanding processes that are strongly guided by context. The chapter discusses various ways in which understanding might be guided by context. A particularly strong form of context-guided understanding is context-driven understanding, a form of which is used in an AI approach to metaphor understanding that the author and colleagues have developed.1

Ever since the discovery of neural networks, there has been a controversy between two modes of information processing. On the one hand, symbolic systems have proven indispensable for our understanding of higher intelligence, especially when cognitive domains like language and reasoning are examined. On the other hand, it is a matter of fact that intelligence resides in the brain, where computation appears to be organized by numerical and statistical principles and where a parallel distributed architecture is appropriate. The present claim is in line with researchers like Paul Smolensky and Peter Gärdenfors and suggests that this controversy can be resolved by a unified theory of cognition – one that integrates both aspects of cognition and assigns the proper roles to symbolic computation and numerical neural computation. The overall goal in this contribution is to discuss formal systems that are suitable for grounding the formal basis for such a unified theory. It is suggested that the instruments of modern logic and model theoretic semantics are appropriate for analyzing certain aspects of dynamical systems like inferring and learning in neural networks. Hence, I suggest that an active dialogue between the traditional symbolic approaches to logic, information and language and the connectionist paradigm is possible and fruitful. An essential component of this dialogue refers to Optimality Theory (OT) – taken as a theory that likewise aims to overcome the gap between symbolic and neuronal systems. In the light of the proposed logical analysis notions like recoverability and bidirection are explained, and likewise the problem of founding a strict constraint hierarchy is discussed. Moreover, a claim is made for developing an "embodied" OT closing the gap between symbolic representation and embodied cognition., ZAS Papers in Linguistics, Bd. 51 (2009): Papers on pragmasemantics

Udgivelsesdato: 16 juli

Although the study of analogy is a common interest for researchers in general systems theory and in artificial intelligence (AI), the research disciplines in these two areas are quite different. In this paper we provide a brief comparison study for these two fields: analogy as the study of systems and analogy as the study of intelligence. We show that these two fields of study are complementary. We also propose the use of logic specification for systems description. A study of analogy that combines perspectives of general systems theory and AI thus becomes evident.

The task of legal research is complex, requiring an understanding of multiple interpretations of legal language, an awareness of the relationships between judicial decisions, and an ability to use analogical reasoning to find relevant documents. Attempts to automate parts of the process with computers have met with limited success. We describe a system called SCALIR which attempts to remedy these problems by combining connectionist and symbolic artificial intelligence approaches. This hybrid representational scheme gives SCALIR the ability to make both associative and deductive inferences. The system also provides an alternative to the traditional view of computer-assisted legal research by using a direct-manipulation style, interface, making searches into an interactive process, and by employing user feedback to improve its performance over time. In addition, we suggest a further application of SCALIR to a part of the analogy task, and argue that this approach is complementary with case-based reasoning techniques.

‘Artificial Intelligence cannot ignore philosophy’ (McCarthy 1988) I shall challenge the claim that good old-fashioned artificial intelligence, or GOFAI (Haugeland 1985) is solipsistic while more recent neural or ‘brain-style’ approaches to AI are not (Rumelhart et al. 1986). After distinguishing GOFAI from connectionism, I will first show that GOFAI is not committed to solipsism but rather to what is more properly called individualism. I argue that the feature of GOFAI which entails individualism is shared by connectionism. Individualism is a metaphysical assumption of both types of AI, one which may indeed be pernicious. It is an assumption which must be located and understood.

Artificial intelligence (including expert systems) (AI/ES) and neural networks (NN) provide methods for formalizing qualitative aspects of business systems. They complement quantitative methods in solving business problems. While AI and NN have the common goal of simulating human intelligence, they use different methods. AI/ES assumes the brain is a black box and imitates the human reasoning process. It processes knowledge sequentially, represents it explicitly, and mostly uses deductive reasoning. Learning takes place outside the system. NN treats the brain as a white box and imitates its structure and function, using a parallel approach to simulate human intelligence. It represents knowledge implicitly within its structure and applies inductive reasoning to process knowledge. Learning takes place within the system. Both AI/ES and NN have great potential to solve qualitative problems, and their integration could provide a powerful tool for dealing with problems outside the domain of current problem-solving methods.

The field of artificial intelligence ( AI) is concerned with improving the understanding of the nature of concepts such as knowledge, intelligence, and consciousness. In the development of the field, modeling the environment has emerged as a central task. Modeling in this sense is a broad term embracing three evolutionary levels: an early stage of biochemical structures, a middle stage in which neural systems emerge, and the later conceptual stage characterized by the use of symbolic structures such as language, interpreted images, and signs. Although AI is particularly concerned with specific area of symbolic knowledge processing, that of logical and relational knowledge, central efforts in this field are also directed at integrating all forms of symbolic modeling into effective solutions for knowledge processing on computers. Thus, methods from computer science are of major importance. This paper illustrates the integration of various forms of modeling within the field of AI. It also shows how ...

This paper examines the epistemological impact of Artificial Intelligence (AI) and rejects some current trends of thought according to which AI should, and will in the future, merge into Cognitive Science (CS). It assesses the role of logic in AI and shows that many arguments directed against "classical" AI are based on a mistaken view of the actual contribution of logic to the field. After a brief discussion of connectionnism, it concludes that both AI and CS have a promising future, that the bridges between them should be reinforced, but that they are likely to remain distinct endeavors.

The theoretical foundations of the logical approach to artificial intelligence are presented. Logical languages are widely used for expressing the declarative knowledge needed in artificial intelligence systems. Symbolic logic also provides a clear semantics for knowledge representation languages and a methodology for analyzing and comparing deductive inference techniques. Several observations gained from experience with the approach are discussed. Finally, we confront some challenging problems for artificial intelligence and describe what is being done in an attempt to solve them.

The present issue is the beginning of a new journal from various sub-disciplines and paradigms in order – Cognitive Systems Research – which we have to construct a coherent picture of how the various developed in response to what we perceive to be an pieces fit together overall. Such a synthesis is unfilled niche in the current literature in the areas of essential to the discovery of designs for general Cognitive Science and Artificial Intelligence. cognitive systems. This is especially important at this early stage of development in understanding cognitive systems. The rationale We feel strongly that the time is ripe for synthesis and integration. First of all, over the last couple of Recent years have seen substantial growth in the years, a significant number of researchers have study of cognitive and neural modeling, cognitive begun to address problems of integrating models, psychology, planning, robotics, language processing, theories, and systems developed within cognitive neural engineering, machine learning, knowledge psychology, linguistics, symbolic and Bayesian artifirepresentation, uncertainty management, artificial cial intelligence, artificial neural networks, evolutionlife, and related areas. Unfortunately, this progress ary computation, and other areas. Numerous recent has also been accompanied by a fragmentation of the edited books, workshops at major international confield into sub-disciplines devoted to studying specific ferences, and special issues of major journals on this aspects of cognitive phenomena often with contradictopic are indicative of the current interest in this tory assumptions, and using what are commonly area. Second, as important as integrating models and perceived to be fundamentally different paradigms theories from different paradigms is the need for (e.g., symbolic artificial intelligence, neural netdevelopment of models and systems that seamlessly works, evolutionary computation, fuzzy logic). The integrate several aspects of cognition and intelligent resulting proliferation of separate journals and largebehavior (e.g., perception, action, language, learning, ly insulated research communities for different topics representation). Some of the recent work in intelliis, in our opinion, an obstacle to progress in this gent agent architectures has begun to address probfield. lems in this area. While work on well-delineated sub-problems (e.g., Despite the growth of research in such directions, planning, cognitive modeling) using specific theoretat present there is no journal that adequately meets ical tools (e.g., search or neural networks) of specific the needs of researchers in this broad field of paradigms does help to make progress on the chosen cognitive systems. A major objective of this journal problems, such efforts must be complemented by is to provide a forum for relatively rapid publication broader theories and models that integrate ideas, of rigorously refereed original research articles and concepts, constructs, tools, and techniques drawn comprehensive surveys in this area, with an em

Learning to drive a car is a long and tedious process. For most of us, it takes many painful hours of training. At the beginning, the car is an external device, reacting in unpredictable ways. But progressively, we get used to some of the dynamics of the complex machine. We have to simultaneously learn basic control operations like switching the gears and pushing the brake, and to combine the operations into higher-level behavioral chunks like starting, stopping, turning, parking, and passing another car. Day after day we get used to the car's space occupation and the time necessary to slow down. The car becomes like our second skin. We learn to adapt each skill to specific contexts such as starting on a steep street, turning on a wet road, passing a long truck or parking in a busy Parisian boulevard. Based on the chunks we can easily anticipate and look ahead. We plan long trips by just focusing on important direction decisions and assuming that all the rest should run smoothly. For those who commute everyday by car, some driving trips are so familiar that the chunks can be done with almost no conscious control. Driving has become as natural as walking: an unconscious experience. However, as in walking, if something unexpected happens (e.g. a child crossing the street and a strange behavior of a car due to a flat tire) we are always ready to take control again and rapidly adapt to the new situation. Tani (2007) doesn't write about cars. However, the article gives precious insights on how neural circuits can perform similar to the process we just described. It shows very convincingly how certain kinds of recurrent neural networks can acquire compositionality without using symbols. Compositionality is the property of systems that can be decomposed into reusable parts. Linguistic systems are typical examples of compositional systems. Discourses are combinations of sentences that are combination of words, which are combination of phonemes. Most sensorimotor behaviors, such as driving, are also intrinsically compositional, reusing basic primitives, chunks or schemas. It seems to be a fundamental feature for achieving complexity. Compositional systems are very common in symbolic artificial intelligence. In these models, each chunk is an explicit structure that can be easily manipulated, organized in hierarchies and used for building plans. Unfortunately, most of the symbolic systems have difficulties in dealing with real-world dynamics for which plans must be constantly reassessed and the chunks must be continuously adapted to deal with the present situation. Tani's neural model not only shows that a dynamical system can exhibit compositional properties without having an explicit notion of chunk or schema, but also convincingly demonstrates that it can deal with complex real-work interaction in real time. The neural network has to discover by itself how to segment the continuous sensorimotor flow into sequences of reusable segments and to simultaneously arrange the chunks into higher-level sequences. The challenge is that both the processes have to run concurrently as in any real-world learning. The cornerstone of Tani's model is that it combines dynamic interactions between top-down look-ahead anticipations and a bottom regression process, permitting to re-situate the system if something goes wrong in the prediction. It is a wonderful illustration of self-organization at work. There is something more. Tani's model tells us something important about the conscious experience associated with such compositional and dynamic management of behavior. Continuing some earlier discussions about the link between this kind of neural models and the conscious appraisal of time (Tani 2004), Tani suggests that when anticipation successively predicts the sensorimotor flow during the execution of a familiar behavioral sequence (e.g. driving on a familiar road), there is nothing to be “conscious” about. However, when a “breakdown” occurs, i.e., when there is a gap between the anticipation of the top-down system and the actual sensorimotor flow (e.g. a child crossing the street), the system has to perform intensive regression, which is an adjustment of the “self”. This bottom-up process may result in the conscious experience associated with the unexpected event. Compositionality and consciousness are likely to be intrinsically linked. The experience of driving a car is made of continuous shifts between unconscious “incorporated” sequences and small moments of awareness. To understand these bottom-up and top-down dynamics, we need plausible neural models of the emergence of compositional behavior. Tani's neurorobotic experiments are important steps in this direction.

This book provides a systematic analysis of many common argumentation schemes and a compendium of 96 schemes. The study of these schemes, or forms of argument that capture stereotypical patterns of human reasoning, is at the core of argumentation research. Surveying all aspects of argumentation schemes from the ground up, the book takes the reader from the elementary exposition in the first chapter to the latest state of the art in the research efforts to formalize and classify the schemes, outlined in the last chapter. It provides a systematic and comprehensive account, with notation suitable for computational applications that increasingly make use of argumentation schemes.

Recent researches reveal that various paradigms of intelligence, such as natural, artificial, machinable, and computational intelligence, can be unified at the logical and functional levels known as abstract intelligence. This paper introduces abstract intelligence as a form of driving force that transfers information into knowledge and behaviors. An architectural framework of abstract intelligence and the generic abstract intelligence mode (GAIM) are formally developed that provide a unified theory for explaining the mechanisms of advanced intelligence. In order to deal with the highly complex and abstract objects in abstract intelligence, denotational mathematics is introduced as a category of expressive mathematical structures for modeling and manipulating high-level mathematical entities beyond numbers and sets, such as abstract objects, complex relations, behavioral information, abstract concepts, knowledge, processes, and systems. Applications of denotational mathematics in abstract intelligence, cognitive informatics, and computational intelligence are elaborated.

Summary form only given. When employing the views of granular computing to the studies of artificial intelligence, it can be seen that there have been various levels of granule existed, system level, element level, and sub-element level, etc., each of which has its own interest of research. As for the system level of granule is concerned, there have also been various views of angle among which structural angle (Structuralism Approach), functional angle (Functionalism Approach) and behavioral angle (Behaviorism Approach) are representatives. Each of the approaches to AI has made progresses and faced problems as well. In this paper, a new approach, named the Mechanism Approach, to the system level of granule in AI research is presented, which focuses on the core mechanism of intelligence formation in general cases. It is discovered that the Mechanism Approach can be implemented via the transformations that are able to converse the information to knowledge and further to intelligence. The new approach is also proven to be able to unify the three approaches mentioned above and thus bring about a unified theory of AI. Both the mechanism approach and the unified theory may make AI research more promising than ever before.

McDermott [12,13] introduced the concept "Artificial Intelligence meets Natural Stupidity". In this paper, we explore how Artificial Intelligence and Symbolic Computation can meet Natural Typography, and how the conventions for expressing mathematics that humans understand can cause us difficulties when designing mechanised systems.

"The AI and HCI communities have often been characterized as having opposing views of how humans and computers should interact" observes Winograd in "Shifting Viewpoints". Reconciling these views requires a thoughtful balancing of assistance and control, of mental and system representations, and of abstract process and contextualized workflow. This workshop examines the gap between HCI and artificial intelligence, with the goal of improving usability of AI systems.

The advent of machine intelligence and the social and ethical issues it raises are examined. The concept of the silicon mind is discussed and landmarks in artificial intelligence (AI) are reviewed. The problem of knowledge representation is considered. The use of artificial intelligence by the military is described. The possible effects of AI on society are also discussed. >

Artificial intelligence and, more recently, neural networks have been claimed to yield revolutionary advances in manufacturing. Jozsef Hatvany had the clearest view of the advantages and problems with AI in manufacturing. The situation with neural networks is similar and can benefit from an extension of Hatvany's analysis.

The current renewal of connectionist techniques using networks of neuron-like units has started to have an influence on cognitive modelling. However, compared with classical artificial intelligence methods, the position of connectionism is still not clear. In this article artificial intelligence and connectionism are systematically compared as cognitive models so as to bring out the advantages and shortcomings of each. The problem of structured representations appears to be particularly important, suggesting likely research directions.

Articicial intelligence is a new science that deals with the representation, automatic acquisition, and use of knowledge. Artificial intelligence programs attempt to emulate human thought processes such as deduction, inference, language, and visual recognition. The goal of artificial intelligence is to make computers more useful for reasoning, planning, acting, and communicating with humans. Development of artificial intelligence applications involves the integration of advanced computer science, psychology, and sometimes robotics. Of the subfields that artificial intelligence can be broken into, the one of most immediate interest to forest management is expert systems. Expert systems involve encoding knowledge usually derived from an expert in a narrow subject area and using this knowledge to mimic his decision making. The knowledge is represented usually in the form of facts and rules, involving symbols such as English words. At the core of these systems is a mechanism that automatically searches for and pieces together the facts and rules necessary to solve a specific problem. Small expert systems can be developed on common microcomputers using existing low-cost commercial expert shells. Shells are general expert systems empty of knowledge. The user merely defines the solution structure and adds the desired knowledge. Larger systems usually require integration with existing forestry data bases and models. Their development requires either the relatively expensive expert system development tool kits or the use of one of the artificial intelligence development languages such as lisp or PROLOG. Large systems are expensive to develop, require a high degree of skill in knowledge engineering and computer science, and can require years of testing and modification before they become operational. Expert systems have a major role in all aspects of Canadian forestry. They can be used in conjunction with conventional process models to add currently lacking expert knowledge or as pure knowledge-based systems to solve problems never before tackled. They can preserve and accumulate forestry knowledge by encoding it. Expert systems allow us to package our forestry knowlege into a transportable and saleable product. They are a means to ensure consistent application of policies and operational procedures. There is a sense of urgency associated with the integration of artificial intelligence tools into Canadian forestry. Canada must awaken to the potential of this technology. Such systems are essential to improve industrial efficiency. A possible spin-off will be a resource knowledge business that can market our forestry knowledge worldwide. If we act decisively, we can easily compete with other countries such as Japan to fill this niche. A consortium of resource companies, provincial resource agencies, universities, and federal government laboratories is required to advance this goal.

Variable-precision logic was introduced by R.S. Michalski and P.H. Winston (Artif. Intell., vol.29, p.121-46, 1986) as a means for representing reasoning in the face of exceptions. Using the framework of belief structures, some of their ideas are extended. In particular, a general framework that easily allows the representation of different types of rules in a unified manner is provided. This approach also allows for a lack of specificity in the associated qualifying probabilities. >

This paper gives a partial overview of what is currently going on in the emerging field known as Artificial Intelligence and Music (AIM). In particular, it focuses on Artificial Intelligence (AI) research areas which, according to a personal view, can significantly influence current music research. The paper highlights how AIM is not a field in which AI technology is merely applied to music. Both music and the AI field can benefit from studies in this discipline: music is a complex experimental domain which can be particularly useful in the formulation and verification of theories about intelligence. This hypothesis is confirmed by the recent works in this field, and particularly by the papers included in this issue of Journal of New Music Research. This paper attempts to set them against an AI global view. The author focuses on the role of some knowledge representation paradigms, such as logic‐based formalisms and network representations. In particular, the role of time in AI formalisms is discu...

This book is reprinted from Artificial Intelligence: An International Journal, Volume 58, Numbers 1--3, 1992. There are reported new results in constraint-based reasoning. Constraint-based reasoning has a long tradition in artificial intelligence. Many problems to which artificial intelligence techniques are often applied can be described as constraint satisfaction problems.

We use the description of the foregoing paradigms for the benefit of an intelligent vs. smart discourse: 1) AS with consciousness would be the result of the development of quantum computers. 2) AS based on logical modules are implemented using our present computers. AS based on expedient recognizers of the outside-world features are implemented with artificial neural networks with chaotic behavior that generates symbols. 4) AS which have been implemented with low-level behavioral modules generate reactivity. The "difference" between intelligent and smart discourses boils down to the fact that intelligent systems are those with the greatest degree of smartness. The ASs proposed here always implies restricted resources for problem solving. Typically, systems or machines are dubbed "smart" when they solve special problems with fewer resources than the brain. Each smart AS is self-organized. We postulate that the combining of paradigms leads to the implementation of artificial systems that are smarter than their components alone. We suspect that this is what the Body Brain does dynamically. The mechanism of the Body Brain makes its smartness superior to that of any other mechanism because it is supported by a large menu of resources combined by the brain on demand.

A fundamental problem in artificial intelligence is that nobody really knows what intelligence is. The problem is especially acute when we need to consider artificial systems which are significantly different to humans. In this paper we approach this problem in the following way: We take a number of well known informal definitions of human intelligence that have been given by experts, and extract their essential features. These are then mathematically formalised to produce a general measure of intelligence for arbitrary machines. We believe that this equation formally captures the concept of machine intelligence in the broadest reasonable sense. We then show how this formal definition is related to the theory of universal optimal learning agents. Finally, we survey the many other tests and definitions of intelligence that have been proposed for machines., 50 gentle pages

Artificial intelligence and physics are two components of 3D games. While they bring the laws of nature and more realism to games, they may be intensive enough to leave a sizable dent on the game 's performance and lead to a serious reduction in fame rate. In this paper, we discuss a game implementation solution with multi-core architectures which hides the computational load of the game intelligence and improves the gaming experience.

Artificial Intelligence, and the cognitivist view of mind on which it is based, represent the last stage of the rationalist tradition in philosophy. This tradition begins when Socrates assumes that intelligence is based on principles and when Plato adds the requirement that these principles must be strict rules, not based on taken-for-granted background understanding. This philosophical position, refined by Hobbes, Descartes and Leibniz, is finally converted into a research program by Herbert Simon and Allen Newell. That research program is now in trouble, so we must return to its source and question Socrates' assumption that intelligence consists in solving problems by following rules, and that one acquires the necessary rules by abstracting them from specific cases. A phenomenological description of skill acquisition suggests that the acquisition of expertise moves in just the opposite direction: from abstract rules to particular cases. This description of expertise accounts for the difficulties that have confronted AI for the last decade.

In this paper we propose a new approach to active video surveillance intelligence systems based on the integration of artificial neural networks (ANN) and symbolic Artificial Intelligence (AI). In particular, the neurosymbolic hybrid system here presented is formed by virtual neural sensors (WiSARD-like systems) and BDI agents. The coupling of virtual neural sensors with symbolic reasoning for interpreting their outputs, makes this approach both very light from the computational and hardware point of view and quite robust in performances.

Artificial Intelligence (AI) and Cognitive Psychology (CP) are two sciences of intelligent systems that share many features. If we want, nevertheless, to contrast AI with CP, we must investigate differences between the strategies they follow in exploring intelligence. To do so, I transform the Turing Test into a more adequate intelligence test based on a necessary condition for intelligence, namely that intensions of second-order intentional predicates are observable in a system (sect. 2). I then contrast CP and AI by their criteria for progress in research on this necessary condition for intelligence (sect. 3).

The attempt to understand intelligence entails building theories and models of brains and minds, both natural as well as artificial. From the earliest writings of India and Greece, this has been a central problem in philosophy. The advent of the digital computer in the 1950’s made this a central concern of computer scientists as well (Turing, 1950). The parallel development of the theory of computation (by John von Neumann, Alan Turing, EmilPost, Alonzo Church, Charles Kleene, Markov and others) provided a new set of tools with which to approach this problem — through analysis, design, and evaluation of computers and programs that exhibit aspects of intelligent behavior — such as the ability to recognize and classify patterns; to reason from premises to logical conclusions; and to learn from experience.

At least three disparate channels have been used to install new knowledge into artificial intelligence systems. The first of these is the programmer channel, through which the knowledge in the system is simply edited to include the desired new knowledge. While this method is often effective, it may not be as efficient as learning directly from environmental interaction. The second channel may be called the linguistic channel, through which knowledge is added by explicitly telling the system facts or commands encoded as strings of quasi-linguistic instructions in some appropriate form. Finally, there is, for want of a better phrase, the learning channel, through which the system learns new knowledge in an inductive way via environmental observations and simple feedback information. These latter two channels are the ones upon which we wish to focus, as they are the hallmarks of instructable systems. Most instructable systems depend upon, or at least heavily favor, one of these two channels for the bulk of their knowledge acquisition. Specifically, symbolic artificial intelligence systems have generally depended upon the explicit use of sentential logical expressions, rules, or productions for the transmission of new knowledge to the system. In contrast, many connectionist network models have relied solely on inductive generalization mechanisms for knowledge creation. There is no apparent reason to believe that this rough dichotomy of technique is necessary, however. Systems which are capable of receiving detailed instruction and also generalizing from experience the both possible and potentially very useful.

Symbolic artificial intelligence is motivated by the hypothesis that symbol manipulation is both necessary and sufficient for intelligence [34]. Symbolic systems have been quite successful, for example, in modeling in-depth natural language processing [[13], [26], [43]], episodic memory [[22], [24], and problem solving [[23], [35], [36]]. In such systems, knowledge is encoded in terms of explicit symbolic structures, and processing is based on handcrafted rules that operate on these structures.

Can machines think? This question has been an object of discussion and debate for over half-a-century. My interest in the question goes back to the beginning of my academic career. Officially, AI was born in l956. At its birth there was widespread expectation that within a few years it will be possible to build machines that could think like humans. I did not share that belief. To the pioneers, symbolic logic was all that was needed. Anything that involved numerical computations was unwelcome. It took close to thirty years for probability theory to gain grudging acceptance. Clearly, adding probability theory to the armamentarium of AI is a step in the right direction. But is it sufficient? In my view, the answer is: No.

Intelligence is a driving force or an ability to acquire and use knowledge and skills, or to inference in problem solving. This keynote lecture describes the taxonomy and nature of intelligence. It analyzes roles of information in the evolution of human intelligence, and the needs for logical abstraction in modeling the brain and natural intelligence. A formal model of intelligence is developed known as the generic intelligence mode (GIM), which provides a foundation to explain the mechanisms of advanced natural intelligence such as thinking, learning, and inferences. A measurement framework of intelligent capability of humans and systems is presented in the forms of intelligent quotient, intelligent equivalence, and intelligent metrics. On the basis of GIM model and theories, the compatibility of nature and machine intelligence is revealed, which forms a theoretical foundation for rigorous study in machine intelligence, AI, and intelligent systems.

Understanding general intelligence and identifying its essential components are key to building next-generation AI systems — systems that are far less expensive, yet significantly more capable. In addition to concentrating on general learning abilities, a fast-track approach should also seek a path of least resistance — one that capitalizes on human engineering strengths and available technology. Sometimes, this involves selecting the AI road less traveled.

Is there an “essence of intelligence” that distinguishes intelligent systems from non-intelligent systems? If there is, then what is it? This chapter suggests an answer to these questions by introducing the ideas behind the NARS (Nonaxiomatic Reasoning System) project. NARS is based on the opinion that the essence of intelligence is the ability to adapt with insufficient knowledge and resources. According to this belief, the author has designed a novel formal logic, and implemented it in a computer system. Such a “logic of intelligence” provides a unified explanation for many cognitive functions of the human mind, and is also concrete enough to guide the actual building of a general purpose “thinking machine”.

Emotions were proven to play an important part in human intelligence. As a fellow of AI (artificial intelligence), there are many issues and argues in the area of artificial emotion (AE). An interesting new model is presented in a musician's point of view to solve the problems how machine can "apperceive", "develop" and "express" feelings emoting like human being. This model simulates the musical communications between musician (e.g. a composer and a pianist performing his works, or a conductor and his orchestra), which seem to be a "pure" emotional exchange.

What is Computational Intelligence (CI) and what are its relations with Artificial Intelligence (AI)? A brief survey of the scope of CI journals and books with ``computational intelligence'' in their title shows that at present it is an umbrella for three core technologies (neural, fuzzy and evolutionary), their applications, and selected fashionable pattern recognition methods. At present CI has no comprehensive foundations and is more a bag of tricks than a solid branch of science. The change of focus from methods to challenging problems is advocated, with CI defined as a part of computer and engineering sciences devoted to solution of non-algoritmizable problems. In this view AI is a part of CI focused on problems related to higher cognitive functions, while the rest of the CI community works on problems related to perception and control, or lower cognitive functions. Grand challenges on both sides of this spectrum are addressed.