…except in the senses not used at this point that everything models itself and, if adapted and everything in the universe is adapted in some way in so far as there are co-origins, models the world sufficiently well. One important point to note is that if there is modeling it is, generally, of the world which includes the machine itself and the designer and builder…

However, a machine that assists in function or functions through copying or simulating understanding or function [adaptation] models or partakes of a model and is an intelligent machine

1.1 Analog vs. Symbolic Machines

An intelligent machine that functions by copying or simulating structure, functioning or adaptation is analog, while one that codes the symbols of understanding is symbolic. A typical digital machine is symbolic – its states are combinations of binary [e.g. 0, 1] states and symbols are assigned combinations that can vary according to application [program.] A typical analog machine models neural structure and, by analogy, is connectionist or associative.

Roughly, the operation of a symbolic machine corresponds to formal understanding as discussed in the division on Knowledge and Action and operation of an analog machine corresponds to the Kantian intuition… and, as was seen in the earlier division, formal and intuitive understanding are identical at root

Symbolic machines are not associative in any direct way and vice versa but according to the Church-Turing Thesis all sufficiently powerful computational devices [with sufficient memory] are universal, i.e. any machine can emulate any other[1]

1.1.1 Terminology

The analog / symbolic distinction also has the following characterizations

Algorithmic computation / signal processing

Classical artificial intelligence / connectionism

AI modeling / neural net modeling

The analog approach is more “bottom-up” and based on a neurobiological metaphor i.e. what are thought to be key properties determining firing of neurons are built in. The symbolic approach is more “top-down” and relies on the science and mathematics of computation

1.1.2 Agents

An agent[2] is understood as:

Perceiving the world [which is the environment and the agent]; perception is thought of as seeking / receiving information about or representation of the environment

Judging; judging is a composite and dynamic activity of processing information to decide actions achieve desirable outcomes; mechanization of judgment includes computation and therefore agency includes computation

The activity is dynamic: in that complete processing of all information is not always possible or necessary or done before acting; what is desirable is contextual and revisable; the processes toward outcomes may be explicitly incremental; processes toward outcomes may be dynamically incremental in that information available before achievement of an outcome may be used to modify decisions regarding actions and what are desirable outcomes

The concept of desirable outcome has the following distinctions: positive outcome sought / negative outcome avoided; intrinsically desirable / rational. An intrinsically desirable outcome would correspond to emotions or drives and for machines they might be “hardwired;” that a goal be associated with a drive or that its satisfaction be hardwired does not make it imperative, for exclusive goals may enter into some kind of competition such as “weighting” or be part of an intrinsic / rational hierarchy. Rational outcomes are explicitly chosen according to some criteria of success including the satisfaction of intrinsic outcomes

Action upon the world

An agent has been defined as rational to the extent that its actions can be expected to achieve its goals or desirable outcomes, from the data given by its perceptual processes. Assuming that an agent might desire [or that the designer, if there is one, might aspire to rational design] to be rational a number of problems arise: there is no a priori reason to suppose that a mechanical agent with fixed goals is a model of being, mind or intelligence or that such an agent will yield the most useful practical devices; there is no a priori guarantee that, even given an infinite amount of time, an agent can determine appropriate actions and this is compounded by the fact that agents in dynamic worlds must operate in real time. In practice, “successful” agents need only be sufficiently successful; where success is context dependent. [Note that the idea of optimality was not even introduced since the concept has significance only in environments that are very simple or definite criteria can be successfully imposed; in such cases the use of single or multiple criteria of optimality may be useful.] Two ontological problems with the concept of rational agency may be identified: the sharp division of activity into perception, judgment and action; and the concept of a being as defined by an interest in goals or desirable outcomes, especially specifiable ones. Nonetheless, the idea of agent as involved in its input-output is an advance over the classical artificial intelligence idea of machine intelligence as performance of isolated reasoning tasks

Any such judgment is value laden for reasoning tasks are useful; the values in question, then, include:

From the point of view of utility: having intelligent machines that operate more or less independently of designers and operators

From the point of view of embedding in the world; having intelligent machines that are less flat from the point of view of organization and so, at the lower levels of organization, tied into the world

And so having machines that approach having mind / being in the senses that animals and humans have mind and being

1.1.3 Computation

So far the concept of computation has received some implicit specification. In order to better perform the following tasks it will be useful to define or specify what is computation

Use the concept of machine intelligence in understanding being / mind

This becomes restrictive on the conjunction of two conditions: first, cognitivism, the idea that the brain is a computer i.e. that at an abstract level the mind is a computer; and, second, on some conceptual or theoretical account of the nature / concept of computation

Developing and simulating machine intelligence for the following purposes: to advance the use and capabilities of machine intelligence; to develop specific applications for use in the projects of the Journey in Being – these are specified in Objectives and include the understanding of being / mind by developing models

In order to develop the concept of computation the following will be useful

Discuss existing concepts of computation, their rationale. Analyze these concepts, understanding how [if] they mesh and what they have in common

Keep in mind that machine intelligence falls under the topic of machines which in turn falls under technology

Remember that the concept of agent is not an alternative to the classical concept of machine intelligence as performance of isolated reasoning tasks but incorporates the classical function into a more independent, “self-sufficient” concept. This independence is not a complete independence – individual beings have degrees of independence and interdependence; in order to take steps to machine being and steps toward more capable intelligent machines it will be useful to examine the kinds and degrees of independence and structure of organisms and individuals. Since the goal of intelligent machine as independent agent / being is only one goal and the use of machines remains important, the performance of tasks remains important

Following are some existing concepts of computation as used in cognitive / computer science c. 2000[3]:

Formal symbol manipulation – here, a symbol is a token for anything that could be a mental content or information such as a concept, representation or word; a symbol is formal to the extent that it is independent of meaning but that the system of symbols belong to a grammar; and there is manipulation to the extent that computation involves a transformation [in time] of the syntactic structures stored in computer memory. Independence from meaning is independence from semantic properties such as reference or truth; the meaning is assigned by the builder / programmer / user

Computation theory based on Alan Turing’s construction of the Turing machine and the related Church-Turing Thesis, above, that all machines [with appropriate conditions] can emulate every machine and so [1] the Church-Turing Thesis is not architecturally limited and, [2] the cognitivist [hypo-]thesis that the mind is a digital computer. The idea of a concept / theory driven notion of computation is [has been] exciting because of the universality of computation [on the Turing notion] and, since the brain appears to be a machine, the applicability of any theory that may be computed including formal logic and computability theory

The concept of computation the process of digital state machines which except for continuity appears to be equivalent to the Turing conception of a computing machine. Infinite state Turing machines including continuous machines can compute [solve] problems unsolvable by finite state machines. However, the concept of a digital state machine is interesting and useful since modern computers are [almost always] digital state machines

Information processing: that comes in semantic, syntactic and practical [Internet…] versions

Classic symbolic architectures: serial, fixed, symbolic, explicit, discrete, high level representations, exemplified by axiomatic inference systems[4]

Connectionist architectures: see comments above, in Analog vs. Symbolic Machines

Practical computation and architecture: the actual projects, commercial, research and theoretical that have driven developments in technology including the consumer technologies and conceptual including AI and quantum computing. Most actual projects are implemented in high-level programming languages such as C++, Java, Fortran. The interaction between this technological richness and cognitive science is rather loose and the future is rather open and unpredictable

The theory driven approach is useful in that theories provide impetus to direction and coherence; as far as theories are reductions the reducibility must be remain in interaction with more basic theories [logic, computability... ,] and with actual developments. Actual development will be an amalgam of theory [concept] driven ideas and actual systems – both academic and research on the one hand and commercial and application oriented on the other hand. Actual developments vary from the theory driven, to practical architecture driven, to the deployment of developed software to theoretical and utilitarian ends

One approach to the nature of computation is as follows

List and describe actual physical architectures; generalize to a system of possible architectures

List and describe actual interpretation based and interpretation free implementations of “mental” function; generalize, using the theory of mental function that includes attitude and action as functions, to a system of possible implementations

Quantum-computation holds the following potential: due to the nature of quantum states, ability to solve otherwise intractable problems [by infinite orders of magnitude;] massively parallel architecture; miniaturization; due to its physical basis, more faithful modeling of brain processing. The following questions are raised: to what extent is the brain a “quantum-computer,” and, to the extent that it is, what is the contribution of the essentially quantum phenomena such as indeterminism and the indefiniteness [or, rather, multiple-definiteness] of quantum states, what is the contribution of the power of quantum computation and what is the contribution of the number of neurons and the massively parallel and connected but classical [non-quantum] structure of the brain

1.1.4 Computational Models of Mind / Cognitivism

From the Church-Turing Thesis the following hypothesis seemed to be a small thought exciting leap: human cognition could be emulated by computers

A variety of views have been held

Cognitivism: the brain is a digital computer. On account of the thesis of cognitivism there is, in cognitivism, a parallel between models of mind and concepts or models of computation

Strong Artificial Intelligence: the mind is a computer program

Weak Artificial Intelligence: the operations of the brain can be simulated on a digital computer

These views have been criticized e.g. John Searle, The Rediscovery of Mind, 1992. Searle held that Weak Artificial Intelligence is trivial and argued against Cognitivism and Strong Artificial Intelligence on the ground that digital computation is syntactic, it is symbol manipulation which is assigned but not intrinsic to the physics. It follows that digital computers, in so far as they are digital computers but not on account of their intrinsic physics, do not have minds and are not conscious

The view, here, is that a computer can be interpreted as a symbol manipulating system but also, especially on the connectionist model, as a physical machine. Or, a posteriori interpretations are at least as significant as a priori ones. Insofar as a computer manipulates symbols, reference, intensionality, meaning, truth are assigned; therefore, as a physical object, without further physical properties, the assignment of semantic properties makes for mind not significantly more than the minimal [if any] mental qualities possessed by a rock. Here, talk of mind refers to high-level mind and not to the mental properties, whatever they may be, possessed by matter merely in virtue of being matter. Therefore, associative machines, on account of their physical properties may come closer to being mental than do serial machines. However, without further development such as low level integration and high level interpretation, the mental properties of associative machines must be small

In so far as machine mind and consciousness is concerned the symbolic view may yet have some promise; however the view of a computer as a co-evolving machine with a posteriori interpretation holds more promise; probably, both the source and recognition of machine mind and consciousness will occur by transference / empathy though Theory of Mind [the theory of how individuals recognize mind in other entities] may also have a contribution

Additionally, there are implications of the introduction of the concept of agent in or as defining machine intelligence. Note that the idea of agent corresponds the inclusion of action as an aspect of mind

1.2 Objectives: Machine Intelligence in Journey in Being

This document is a specification and design of related uses of intelligent machines [machine intelligence] in Journey in Being:

1.2.1 Understand and Construct Being / Mind

Design, experiment and construction of a machine, likely an [intelligent] agent, to simulate or have being / mind contributes in a number of ways:

As an entity in the Journey itself; in the division Knowledge and Action, the questions of possibility, probability and value of transformations were considered. Computation and machine agency are regarded as one approach to magnifying the probabilities of transformation. The full implementation is a future and ambitious project. However, even as a tool, there is some contribution to “being.”

As a data point: if human and animal intelligence / being is one data point, machine intelligence is an additional one

The design and construction requires understanding of being / mind. Use of the understanding in a design and experimental setting will contribute to the understanding

Artificial life

1.2.2 Assistants and Independent Tools in Research and Other Tasks

Conceptual representation, understanding and research: as an example, I compared database representations of two formulations [one based in subjective idealism and the other in materialism] of Evolution and Design to determine which topics were of fundamental importance. I have used standard software to develop a number of minor applications e.g. to assist in determining the main topics of Journey in Being

Technical applications

Text and website production

Management

1.2.3 Design Principles

These design principles are for design / use of computation in the objectives, above

The first set of design principles for the use of machine intelligence in Journey in Being are stated as a set of oppositions. In the following each side of the opposition has importance

Theory and actual objectives

Machine as assistive and independent tool

Use as tool leads to independent deployment

Reasoning tasks and machine as agent

Design and [co-]evolution

Assigned and intrinsic reference

Flat and or multi-leveled with regard to organization

Lowest levels designed or lowest level integral with nature

Human / machine interaction will include dynamic interaction; i.e. machines are designed but, additionally, may interact in co-evolution in which [1] both human and machine may change and, [2] through design and selection [rational and evolutionary] change may occur by transference of properties / characteristics. The dynamic loop is: model – performance – redesign – new model

Non-design effects

Experiment

Hardware and software

Software description assigned or interpreted

Digital and or connectionist architecture

The following are also important

Foundation in mind / body and knowledge / being considerations of the earlier division Knowledge and Action

Knowledge representation [concepts] dynamics and foundation in the earlier division

Integrating the digital and associative architectural models of cognition

On account of the parallel between mind and computation in cognitivism, computer architecture, outlined in detail in Computation, provides models of mind i.e. of cognitive architecture. Roughly cognitive architectures come under two classes: serial or von Neumann and connectionist or associative architectures. The main idea of the von Neumann architecture is for programs and data to be stored together i.e. from an abstract point of view, there is no distinction between program and data. The idea of the multipurpose programmable machine is implicit in the idea of storing programs. The main units of the von Neumann architecture are processing, memory, input and output. Cognitive architectures inspired by the von Neumann architecture of computation begin with the early idea of the production system of Herbert Simon and Alan Newell; the first general model of a production system was the General Problem Solver in which the “structures” represented are primarily resident, in principle, in the program. Input / output information processing theory is another class of cognitive architectures derived from the von Neumann concept. In representational models, the structure that is modeled is put in explicitly in the data rather than implicitly in the program. In associative architectures, processing is done by many sub-processors operating at the same time – in parallel; these date back, at least, to Aristotle’s theory of memory and include the work of McCullough and Pitts in building models of neural networks and Donald O. Hebb in explaining psychological phenomena from global neurophysiological models. Since the work of McCullough and Pitts and of Hebb in the 1940s, various theoretical developments, see Cognitive Architecture[5], a variety of theoretical developments has provided a foundation for and introduced tools for application of associative architectures. And, finally there is a variety of hybrid von Neumann / associative architectures that combine the strengths of the two approaches – the ability of von Neumann architectures to match human level competence and the ability of associative architectures to model the context specific and multi-tasking aspects of human intelligence. Finally, the fact that, in the brain, there is no intrinsic distinction or separation of architectures and that high level architecture / processing is built from the low level points to a limitation and opportunity for progress in the concepts of cognitive architectures

The von Neumann architecture inspired the concept of the programmable computer as a general purpose machine. The introduction of compilers that translate a program written in a high level language that is not machine specific in to a machine specific low level program significantly enhanced the use of the computer as a general purpose machine. Specifically, not only is it possible to program machine intelligence [as part of the performance of agents,] it is also possible to model agent and environment, i.e. to model agents and there actions and performance

1.3 Theoretical and Conceptual Background

A significant degree of useful theory has been covered in the present and other divisions. Therefore, the following treatment will be brief; it will primarily make reference to the locations of the developments. The primary locations are the discussions of knowledge, concepts, language, metaphysics, mind, mind / body and cosmology, especially the theory of origins and evolution in the division Knowledge and Action; and the discussions of being and of technology and machines of the present division

1.3.1 Ontology

The following topics are useful:

Can a machine have life, mind or being? Here, it is implicit that life, mind or being refer to life, mind and being as we know them and not “primal” mental elements or being. The primary reason for this specification is that due to the nature of life, mind, and being as slack concepts the answer without such specification must be “yes.” It is further implicit that a machine is something that is built from any natural elements and is not restricted to the classical idea of a deterministic machine. One answer to the restricted question is that, since the brain or body is a machine, it is obvious that a machine can have mind, life, being! However, if by machine we mean something designed and built by agents that have mind, life and being to perform a specified function, the answer is not clear. Specifically, it is not clear that the performance of the function is the equivalent of having mind and so on

What is it about living beings that makes for possession of mind? Note the following possibilities: multiple layers of organization from atoms / particles up; deep grounding and adaptation of all layers; every cell in an organism is a variation of a single cell; every cell, with exceptions, contains the entire genetic code

What is the nature of tools and machines; tools and machines as dynamic extensions vs. independent agents

The question of how to recognize mind is important

One approach is theoretical: to use the nature of mind and its characteristics to determine whether a machine, in virtue of its function, possesses the specified characteristics. Note, that since there is mental content input-output replication does note make for mind; therefore, function must be interpreted to include description of internal elements and processes. Problems with this approach include that mind is not a definite concept as noted in Dimensions of Mind / Being: Introduction; this, however, enhance recognition

Another, related, theoretical approach is what is called Theory of Mind i.e. understanding of how individuals recognize the presence of mind, life or being in other individuals. The so-called Turing test is a sort of primitive test; there is no reason that the “machine” should be hidden from the “observer” who only sees output from the machine; at the same time the observer need not be exposed to all internal details. The Turing test may be interpreted to rely on perceptual similarity: comparison, empathy and intuition. It may be argued, with some validity, that recognition of mental phenomena in others is automatic; however, this is not true for all mental phenomena: some mental phenomena are invisible and some may be purposely concealed. Thus, a machine that had the qualities and degree of human mind ought to be able to deceive individuals and machines into thinking that the machine does not have those qualities. There is also the issue of recognition in the case of other kinds of organism and in machines. When the idea of recognition is extended to conceptual similarity, the test may be interpreted as the theoretical approaches mentioned earlier

When organisms and machines co-evolve, recognition will probably be automatic and not require theory. At the same time, theory will be useful

1.3.2 Knowledge

The following topics are useful:

Concepts[6]

In concepts, the idea of concept was introduced. It was noted there that concepts may receive over-articulation; however, it was also noted that the proper degree of articulation is relative to the purpose. Here, it is appropriate to further analyze the idea of concept. The purpose of the analysis is for use in machine intelligence and, since machine intelligence occurs in interaction with and is modeled on organic intelligence, the study of concepts as they occur in both organisms and machine / computational systems is important

Concepts have narrow content in an individual’s representation of the world and broad content as a symbolic element of interaction among individuals and the world

The intension of a concept is its meaning and the extension is the set of objects to which the concept may refer. These are rough definitions that may be improved along a number of dimensions. I say that intension and extension are mutually determining over time; as a result a specific intension and extension is, in general, never arrived despite the feeling that concepts are innate. Concepts are constructed; use becomes automatic; the origins of a linguistic concept may be in the phase of pre-language; what is a construct functions, within limits, as and feels as though it is a priori

In machine intelligence, one use of concepts is as symbolic elements from which representations of knowledge or an area of expertise is constructed; in a connectionist machine a concept would be identified [rather than defined] as an internal pattern that frequently arises in a variety of situations

In psychology, three main traditions of research are cognitive development originating with Piaget, classifying the world into categories in behaviorism, and lexical semantics the study of concepts through words that are commonly used in their expression. For further information see the references cited in the footnotes. Further information will be included in the present document dynamically as need arises in connection with development of research tools and other applications

Knowledge representation[7]

Knowledge representation arises in the coding and use of areas of knowledge, sometimes expertise, for the purposes of use. Practical uses are expert systems and theoretical, later, research uses include conceptual synthesis and concept recognition. I have developed some applications of the latter type as described below in the applications

Knowledge representation requires: a vocabulary or set of terms and the related concepts that define the field covered; this system is called the ontology. Both philosophical and practical aspects of the ontology are important; however, the development of practical systems may require, at least temporary, suppression of philosophical concerns such as faithfulness, consistency, completeness. To a degree such concerns are introduced via defined syntax and interpretation or semantics. The use of knowledge representation involves input / output computation such as: initial / final conditions; problem statement / solution. The computation requires modeling. Depending on the context models may be more or less realistic and general: logical, diagrammatic, dynamic in the sense of physics and so on. The greater the variety and realism or expressive power, the greater is the deductive complexity and computational resources are limiting in this regard. One problem of expression is the distinction of dynamic from contextual elements that is automatic for an organism but not at all for a machine. Such problems of knowledge representation may be addressed by the introduction of realism in the cognitive process of the knowledge representation and dynamics; and by synthesis of such realism with knowledge representation. One approach is the use of connectionist models that use connectionist architectures to model cognition. “Situated theories” use models of emergence of behavior as a consequence of interaction with the world

In commercial applications, database technology is used as the basis of knowledge representation; the ontology can be incorporated as data e.g. as a set of propositions and or as code e.g. as methods of inference translated into algorithms

1.4 Tools and Tasks

Basis and definition e.g. the forms of information correspond to the forms of human intuition, formal representation and perception

1.4.1 Tools

Topics in Computer Science[8] and Artificial Intelligence[9]. Knowledge and conceptual tools: networks and libraries; dictionaries and encyclopedias; texts; knowledge bases and representation – trees / databases, question and answer; reference systems – informal through tertiary and search tools

Hardware and environment issues [10]

1.4.2 Tasks

A System of Tasks[11]

Document management: storage, production, editing, publication: Internet, paper. Text system management: production, edit, update, multiple concept representation and translation, index and contents; general symbolic processors

Intelligent assistance: assistant, interactive or dynamic, independent. Research: group process, communication, projects; funding, support; management and administration; communication

Management: personal, enterprise

1.5 Implementation of Objectives

The present implementation of the objectives is:

1.5.1 Applications

Applications so far

The computer as environment: comparison with books and writing, leafing through electronic documents and search; learning and intuition; use of massive computational storage, processing and representational capabilities; graphic interfaces: interactive, intelligent, customizable; an interface for communication; file systems, their intuitive / logical structure; ubiquitous presence; automation; number codes and concepts; screen presence – large / multiple screens and workspaces, partitioning, cut and paste; draft vs. production formats; document navigation ; standard software as meaning processors

Text / concept system: production, update and maintenance and publication; complex documents, propagating effects of changes; linking; generic outlines and connectivity; text production and knowledge base templates – see links to the applications general, knowledge base and text production templates; specific systems: journey in being, core curricula for science, engineering, and humanities; knowledge applications, encyclopedia, encyclopedia as database, articles listing, conceptual and coded listing; human knowledge project; idealism / materialism databases of Evolution and Design – see link to the application; scratchpad and conceptual experimentation

1.5.2 Plan

Plans are classed according to the Objectives: Machine Intelligence in Journey in Being [construction, transformation; the understanding that comes from consideration of and actual construction and transformation; and application] and Theoretical and Conceptual Background

Plans include

Human Knowledge Project; see the document, Design for a Journey in Being

Use or deploy existing and commercial applications, then develop; exception: learning

Assistants, then interactive systems, then emulators and agents

Intelligent capability for defined application, then imitation and simulation of intelligent life

Machine as life, as being, as having mind, as conscious

Evolution and life; life design - a concept

[1] There is an issue of whether semantics is realized in computation. In proof theory, within limits, the semantic relations between propositions are isomorphic to syntactic relations between sentences; thus if the syntax is intrinsic so is semantic

[2] I have referred to the article Computation by Brian Cantwell Smith in the MIT Encyclopedia of the Cognitive Sciences, 2^nd edition, 1999, Robert A. Wilson and Frank C. Keil, eds. However, in the present essay, there are a number of differences in treatment and emphasis

[3] Computation by Brian Cantwell Smith in the MIT Encyclopedia of the Cognitive Sciences, 2^nd edition, 1999, Robert A. Wilson and Frank C. Keil, eds.

[4] Computation by Brian Cantwell Smith in the MIT Encyclopedia of the Cognitive Sciences, 2^nd edition, 1999, Robert A. Wilson and Frank C. Keil, eds.

[5] Steven Sloman in the MIT Encyclopedia of the Cognitive Sciences, 2^nd edition, 1999, Robert A. Wilson and Frank C. Keil, eds.

[6] Concepts by James A. Hampton in the MIT Encyclopedia of the Cognitive Sciences, 2^nd edition, 1999, Robert A. Wilson and Frank C. Keil, eds.

[7] Knowledge Representation by Patrick Hayes in the MIT Encyclopedia of the Cognitive Sciences, 2^nd edition, 1999, Robert A. Wilson and Frank C. Keil, eds.

[8] Topics in computer science. Hardware systems, artificial intelligence , numerical analysis, software systems, mathematical foundations of computing, analysis of algorithms, typography and computational models of language, general interest

[9] Topics in artificial intelligence. General: production systems, neural networks, genetic algorithms and programming, computer vision, search including heuristic search, planning, logic, knowledge representation and reasoning, managing / reasoning under uncertainty, common sense reasoning in logic, Bayes networks, automatic planning and multi-agent communication, robotics and computer vision, machine learning, intelligent architectures - connectionist or associative vs. von Neumann, and natural language understanding and processing, interactivity, narrative, and artificial intelligence. Knowledge Representation: declarative knowledge representation methods. Time and action, non-monotonic logics, causality, inheritance and description logics, ontologies, contexts, knowledge acquisition and reformulation, multiple views, abstraction, deduction vs. abduction, knowledge and other mental attitudes. History: knowledge (declarative) vs. procedure based systems. Knowledge-based systems and applications: knowledge-based (expert) system technology is the most widely-used application technology to emerge from AI. Topics: basics of knowledge based systems (KBS) and expert systems (ES); technology transfer from research to industry; knowledge engineering, KB programming, knowledge acquisition methodology; evolution of the technology as applied to business and government problems, current and future impact. Robotics and computer vision: manipulator kinematics and inverse kinematics; manipulator dynamics, motion, and force control; motion planning and robot programming. Robot programming topics include: basics of motor control and sensor characteristics; sensor fusion, model construction, and robust estimation; control regimes (fuzzy control and potential fields); active perception; reactive planning architectures; various topics in sensor-based control, including vision-guided navigation. Some increasingly complex behaviors for mobile robots: simple dead reckoning and reactivity, planning and map building, communication and cooperation. Issues and techniques of computer vision. Image formation, edge detection and image segmentation, stereo, motion, shape representation, recognition

[10] Hardware classes and environment issues. Hardware. Computation: processors, memory; storage; input / capture: keyboard, connectors, microphone / audio, camera, video; output / display: monitor, projectors, speakers, printers. Environment issues. Conditions: humidity and wetness, marine and submerged; temperature... polar, desert; atmospheric: caustic, corrosive, dust and particulates, pressure extremes; space; industrial; acceleration and shock. Mobile vs. stationary factors: size, area, weight; mounting and carriage; modularity – auto install; power – internal / external, local / imported; durability; communications

[11] Software classes. Systems and communication. Systems Software: 1. Applications development system: development: languages, compilers, visual environments, cross compilers, debugging and testing; 2. Operating systems: processor control, memory management, file management and linking; user interface – GUI; network interface; device drivers. 3. Utilities. Communication Software: 1. Data communication: network management and operating systems; data compression; Internet; 2. Telecommunications E-mail, fax, message center, video/teleconference; 3 Security and encryption. Applications software. Production Level Data Processing these have basis in human modes of perception, meaning and communication: word / text / font processors – general, scientific, other; document management and linking; publishing, desktop publishing, typesetting; numeric, array and spreadsheet; database management; graphic – draw, paint and photo, converters, general / scientific presentation, flow sheet / art libraries; CAD; video, sound, music and speech – voice recognition, scanning and handwriting software, video and digital camera… and synthesis, musical instrument digital interface ; multimedia - playback, production and development. Knowledge and Knowledge Systems, Artificial Intelligence: Data / knowledge bases; AI / expert system, AI based design and planning tools: genetic algorithms, neural nets. Specialized applications: Design, decision, planning and government. Science: mathematics and statistics; science and engineering. Arts and humanities. Education: interactive classroom simulation, computer based training, text and media development. Law. Medicine: medical, psychiatric and management - managed care planning. Commerce: Business and finance, industry and agriculture, service and service industries, and trade. Grants