Appendix A: One Hundred Definitions of AI

Transcription

1 Appendix A: One Hundred Definitions of AI Massimo Negrotti The following is a list of 100 definitions of AI collected in our first survey of AI people, during the 8th I}CAI held in Karlsruhe in The AI researchers and designers interviewed came from many countries, and particularly from the USA, Canada, Great Britain, West Germany, France, Italy, Austria, Switzerland, Holland, Sweden, Belgium, and Denmark. Notice the extreme variety of definitions, ranging from the technologically oriented to the pure-research oriented; from the intellectually committed to the comical. Ability of machines to adequately function in human culture Ability of machines to do something which looks like an intelligent behaviour AI gives reasonable answers to ill-stated problems based on inaccurate knowledge AI is always beyond human intelligence AI is getting more out of a computer then the designer could predict AI is that which machines cannot do - once a problem is solved it isn't AI AI is the attempt to map human intelligence on to computer software Applied epistemology Art of making computers think and learn Attempting to build theories that help to build machines that mimic behaviour that humans call intelligent Attempting to compete with a human being without feeling Attempting to find useful solutions to problems which we do not know how to solve Attempting to pass a sophisticated version of the Turing test Attempting to reinvent human abilities on a different technological basis Attempting to represent human thought processes Attempting to reproduce aspects of human intelligence in machines Attempting to simulate human thought processes on alternate hardware Attempting to understand and reproduce man's intellectual processes Attempting to understand the processes of consciousness

2 156 Understanding the Artificial: On the Future Shape of Artificial Intelligence Attempting to write programs that exhibit intelligence in their results and their methodological criteria Attempting to explain human behaviour from a design stance Behaviour which if observed in a human would be ascribed as intelligent Building computer programs which behave intelligently Building understandable programs that can be improved "externally" Cognitive mechanics Computational modelling of cognitive processes Computational study of learning, reasoning, problem solving and understanding Computer simulation of human behaviour in order to produce useful and if possible illuminating programs Computer simulation of intelligent processes (often human) Constructing theories and models of cognitive processes and implementing them in experimental systems Construction of finite processes which duplicate or mimic human behaviour Construction of machines which behave in a way people describe as intelligent Creating a theory of intelligence in abstracto - regardless of carrying agent Creation of systems which look intelligent to human beings Cybernetics with a program Decisions and work with incomplete information Developing models of intelligent processes Development of a theory of mental processes as implementable procedures Development of formal models of human behaviour Development of judgemental heuristics Doing with computers things requiring intelligence in humans Enabling an automaton to perceive process and manipulate complex data Engineering of conscious behaviour Ensemble of decisions and actions which permits a system to survive Exploring what cannot be algorithmically understood now in human affect reasoning Extension of human capabilities in the area of knowledge processing and problem solving Formal and operationalised model building Formal models of complex and context-sensitive information processing Formalisation of intelligent behaviour Getting computers to act as if they are doing intelligent things How to accomplish tasks by exactly defined processes which humans call intelligent Information sculpture. Left-overs other computers systems can't handle Intelligence comparable to human for task Learn how people think and how to do it better using computers as tools Learned or developed behaviour Make a computer simulate intelligent human behaviour Make programs behave in an intelligent way using knowledge Making computers more usable Making machines behave in a way functionally equivalent to humans in specific domains Making machines from the models for our own minds Management of knowledge with computers

3 Appendix A: One Hundred Defintions of AI 157 Mechanical epistemology? Model the mind brain No matter how "intelligent" AI programs become, they will never be better (in a moral sense) than the people who use them Performance of tasks previously considered as exclusive to human intelligence Produce artificial machines who acts or behave as humans Producing a formal theory of cognition Production of models that perform tasks normally said to require intellect and! or a complex neural system Programming machines to do complex tasks using mind as model if necessary Reasoning machine learning Sagacity Science aimed at producing systems that perform tasks as a human applied epistemology Science of intelligence Science of trying to build automata which behave intelligently Select a good strategy for performing a task while creating new ones when needed Simulation by computer of intellectual tasks considered typical of human intelligence Simulation of human behaviour by a machine Simulation of human intelligence Simulation of mental processes Simulation of problem solving processes in humans Simulation of purposeful behaviour Simulation of the functions of mind Solving the riddle of human with satisfactory methods Strengthen human intelligence by computers Study of apparently or actually intelligent behaviour in machines Study of control of understanding Study of general principles underlying mental processes Study of how humans (and animals) actually do the amazing range of activities that we perform: in particular intelligent activities. The building of mechanisms that exhibit aspects of human behaviour; usually intelligent Study of intelligence by building and analysing computer programs Successful problem solving on "difficult" tasks Symbolic processing To get machines to perform complex tasks which require intelligence from people Techniques which make the computer act as a human being The step just beyond where AI research has gotten to today (cynical!) Theory of entities behaving in a given environment achieving goals There is no adequate description To get a handle on mentality To me it is the study of the nature of intelligence independent of (human) hardware Trying to achieve a closer match between computers and human thought behaviour Trying to make computers do things that people can do better

4 Appendix B: An Attempt at Getting a Basis for a Rational Definition of the Artificial Massimo Negrotti Natural system Artificial device Fig. B.1. The artificial as an extraction of a sub-system from a natural system in order to reproduce some functions of it. Note: The number of homologous components is reduced. The homologous components are structurally different. New components may appear (such as F). Some kinds of internal or external relations are lost, and others may be added. New types of internal and external relations may appear. Some basic statements: 1. An artificial device is a machine which reproduces some essential functions of a sub-system of a natural system. 2. In the artificial sub-system the number of homologous components is reduced. 3. In the artificial sub-system the homologous components are structurally different. 4. In the artificial sub-system new components may appear. 5. In the artificial sub-system new types of internal and external relations may appear. 6. In the artificial sub-system, some kinds of internal or external relations are lost, and others may be added. 7. Every artificial device is a machine, but not all machines are to be conceived as artificial devices.

5 160 Understanding the Artificial: On the Future Shape of Artificial Intelligence 8. The performances of an artificial device usually show a different spectrum (sometimes wider and sometimes narrower) compared to the one shown by the correspondent natural sub-system. 9. The research and the development of enhancements of the artificial device consist in the deepening of its own artificial characteristics as such and, usually, this moves the new generations of the device further and further from the natural sub-system. to. The artificial device will be accepted as a good reproduction of the natural system if, and only if, its functioning allows a good reproduction of the main and essential features and performances of the natural sub- system.

6 Subject Index Acceptance rules 27-8 Adverbs, temporal 24, 25 AI and biological intelligence and culture (r8 applied (AAI) 79 as dialectic of science and technology as first phase in process of externalisation 123 as new field of research 66 as technological field 66 assessing achievements of 63 basic assumptions and theses 55-6 claim to legitimacy 61-3 claims of 60 classical symbol-based. See Symbol system approach consequences of generally recognised failure of 62 cultural roots 1-11, 61-3 current approaches to research 2 definitions of epistemological status 57 future 1-11, 63 history of 33, 45 nature of projects new directions for research (r8 new way of conceiving 56 novelty of 61 original programme for 57 progress of 2 reliability of 8-10 Searle and Dreyfus' arguments self-limiting redefinition of 62 strong 60,64,127,128 weak 60, 64, 127 ALIBI 84 Aplysia 95, 99 Applied artificial intelligence (AAI) 79 Artificial device Artificial heart 7 Artificial intelligence. See AI Artificial Intelligence Project 146 Artificial neural nets 37 Artificial sub-system Artificiality definition of 9 essence of7 need to understand 1-2 Aspect 19 Back-propagation of errors 37 Biochemistry 15 Biological intelligence and AI Bipolarity 77-9 Blackboard organisation 81-2 Blackboard representation 86 Brain and individual variability modelling 99 modelling versus mind making Calculating machines 33 Calculemus 140 Calculus ratiocinator 140 Callidiority 79

7 162 Understanding the Artificial: On the Future Shape of Artificial Intelligence Cartesian tradition 41 Categorisation 98 Cerebral plasticity 93 Circumspection 42 COBOL 79 Cognitive approach, perspectives of 29 Cognitive processes, models of 146 Cognitive project Cognitive science Cognitive Science Society 45 Cognitivism Combinatorial explosion 36, 37 Common sense 13,139,144,151 Common-sense knowledge 44, 45, 47, 82 Communication processes 8 Compositional rule 26 Computational simulation 135 Computer intelligence 56 Computer models 58 Computer programs 71, 102, 104-5, 127, 133 Computer science 14-15, 60-1, 62, 77 Computer simulation 98 Computer systems 14,33,78 Computer technology 69,77,101-2,121 Connection machines 100 Connectionist approach 3, 150, 153 Cybernetic society 1 Cybernetics 57 renewed interest in 3-6 theory 4 Data structures 44 Decision support systems 80 Default assignments 41 Deictic adverbs representation of 25 rule for introducing 26 De-skilling 80 Digital computers 33-6 see also Computer Discursive Representations Theory (DRT) 19,29 Distributed architectures 8 Distributed control 4 Dream memories 97-8 Dualism 3,5,7 Ecology of technological environment 2 Effectuation 18 Environmental input 93 Epigenesis and plasticity Etruscan 85-6 Etruscology 85-6 Expert behaviour, meta-models of 85-6 Expert refueller 81 Expert systems 3,9,58,63,65, 79-81, 83, 118, 141, 146, 153 reliability 9 Externalisation and musical notation 122 concept of 122 of knowledge. See Knowledge externalisation of language phenomenon of 122 Facts as primitive objects 34 Family relationships 46 Fault-toleran behaviour 8 Frame, definition 41 Gamba machines 38 General intelligence 149 General Problem Solver (GPS) 43,142-3 General systems theory 1 Generalisation 98 Gestalt misconception 37 Gestalt thinking 38 Global approach Gregorian chants 106,120 Heterogeneity concept 2 Heuristic problem solving 35 Heuristic refueller 80 Hierarchical models 86 Holism 38 versus localisationism 96-7 Housing benefit claims 105 Human behaviour 134 Human-computer interaction (HCI) studies 9 Human intelligence 5-7, 70, 149, 150 Human reasoning 71 Human vision 134 Ideas Pertaining to a Pure Phenomenology 41 IF-THEN process 71 Implementability principle 59 Industrial Revolution 102 Information processing 132, 133, 152 Information recording 95-6 Information society 124 Information technology cultural history of 102 demonstration of new 108 epochs of Integrated approach Intelligence definition of 47 residual 10 Intelligent behaviour 7, 34, 35,42,46, 59, 64,72,91-2,139-47

8 Subject Index Intelligent machines and intelligent fragmentation mechanical model of Intelligent system 15 Interfaces 146 Intuition 5 Knowledge evolution of 116 new type of 13 Knowledge base 43, 80-3 Knowledge engineering 65, 73, 80, 81, 85, 86 and elusive answers 118 Knowledge externalisation and progress and social organisation of specialists 103 as historical process epochs of 103 through computer programming through writing made for human reading Knowledge representation 44 Knowledge sources 82 Language externalisation Learning experiments 53 Learning machines 37 Learning process 73 Lexical database 83 Linguistic structures 16 Localisation in time 19, 20 Localisationism versus holism Look-up table 73 Machine intelligence 149 MACSYMA143 Mainstreaming 77, Man-machine sub-systems 2 Matching-based decision making 72 Memory approach in AI 98 as categorisation procedure 98-9 microlocalisationist view of 99 see also Dream memories Mental processes 67, 69, 71, 132, 152 Meta-models of expert behaviour Mind making versus brain modelling Modular systems 8 Monism 3, 5 Multilayer systems 37, 38, 46 Musical notation and process of externalisation 122 as case of co-evolution as creator of wrong notes as test case 105 complexity in earliest systems of 106 evolution 109 9th to 17th century th century th century th century th century th century th century 114 polyphonic music standardised scale Natural intelligence 101 Natural language logical semantics of 16 programming in 18 semantics 13 temporal constructions in 19 temporal structures of 29 understanding versus programming languages 101 Natural language processing 7,83-4 cognitive dimension in Natural language understanding system, functional architecture of 15 Nature of Human Intelligence, The 143 Negative feedback loops 154 Neo-cybernetic studies 3 Nervous system, plasticity of 92 Neural networks 3, 46 artificial 37 limited inference capability of 5 modelling 47, 48 tradition 4 Neurobiological doctrines 4 Neurolinguistics 15 Neuronal machines 99 Neuronal plasticity Neuronets 58, 72 Neurophysiology IS, 133 Neuropsycholinguistics 18 Neuroscience 35 Non-perturbed intelligence 9, 150 Norwegian Central Office for Social Security 105 Office automation 80 Omega point 79 ONOMATURGE project 83-4 Organisation theory 2 Parallel computing 16, 18 Parallel Distributed Processing 45

9 164 Understanding the Artificial: On the Future Shape of Artificial Intelligence Parallel distributed processing (PDP) 49-50,98 Parallel processing 4 Pattern-directed inference systems (POlS) 80 Pattern recognition 5,37,39 Perceptron 36-8,45,53 Perturbed intelligence 150 Phenomenology 40, 43 Philosophical premises 4 Phonology 135 PhYSical symbol system approach. See Symbol system approach Plasticity and epigenesis 93-4 as repairing phenomenon 94-5 biological concept of 92 cerebral 93 for the evolutionist 92-3 neuronal 95-6 Pragmatics 15 Primary regulation 4, 5 Probabilistic learning algorithm 37 Problem-solving techniques 7, 79 PROLOG 16,19 Purpose (in)dependency of representation 83-4 Quantity, challenge of 81-2 RAFFAELLO 84 Reasoning, automatising 19 Redundancy principle 8 Reliability AI 8-10 concept of 8-10 expert systems 9 measurement 8 software and hardware 8 Retina 49 Robotics 58,146 Rosenblatt-Minsky polemics 5 Rule-based information process 56 Rules 130-1, 135, 136, 144 Science and technology 78 Secondary regulation 4 Self-organising machines 8 Semantics 15, 135, 136 of time in language Semiotic systems 102,104, 109, 114, 115 Sequence analysers 98 SHRDLU 43, 44 Simulation tradition 9 Skill requirements 80 Social organisation of specialists 103 Social science Social structures 86 Spatio-temporal reasoning, automatising of 14 Stepwise problem solving 47 Strategic Defense Initiative (SOl) 8 Structural primitives 40 Symbol system approach 3-5, 34, 35, 38-40,43,47,57,150 Symbolic manipulation programs 5 Systematic approach 145 Systems dynamics 1-2 Systems modularity 8 Tape recorder Technological environment 1, 2 Temporal adverbs 24, 25 Temporal connectors 26-7 Quand27 Temporal language 24 Temporal logic 23,29 Temporal markers in language 20 Temporal structure 20 of natural language 29 rules for constructing 28-9 Thinking machines 3, 139 Time adverbs of 24, 25 in language elements of expressing semantics of localisation in 19, 20 Tractatus Logico-Philosophicus 34, 40 Tradition 5 Transcendental phenomenology 41 Turing test 70, 129, 142, 143, 146 Universal heuristics 143 Universal machine 140 User role 9 Verbs lexical semantics of 21-2 semantics of tenses adverbs and connectors 25-9 formal and linguistic framework 23-5 Visual cortex 94 Visual recognition 98 What Computers Can't Do 121 Word formation 83-4 World as totality of logically independent atomic facts 34