From rutgers!sun-barr!cs.utexas.edu!uunet!mcvax!ukc!icdoc!syma!aarons Mon Aug 7 12:31:15 EDT 1989 Article 4702 of comp.ai: Path: sunybcs!rutgers!sun-barr!cs.utexas.edu!uunet!mcvax!ukc!icdoc!syma!aarons >From: aarons@syma.sussex.ac.uk (Aaron Sloman) Newsgroups: comp.ai Subject: Re: Is there a definition of AI? Keywords: defining AI Date: 6 Aug 89 17:02:11 GMT References: Organization: School of Cognitive & Computing Sciences, Sussex Univ. UK Lines: 199 kim@watsup.waterloo.edu (T. Kim Nguyen) writes: > Date: 5 Aug 89 02:17:40 GMT > Organization: PAMI Group, U. of Waterloo, Ontario > > Anyone seen any mind-blowing (I mean, *GOOD*) definitions of AI? All > the books seem to gloss over it... > -- > Kim Nguyen kim@watsup.waterloo.edu > Systems Design Engineering -- University of Waterloo, Ontario, Canada Most people who attempt to define AI give limited definitions based on ignorance of the breadth of the field. E.g. people who know nothing about work on computer vision, speech, or robotics often define AI as if it were all about expert systems. (I even once saw an attempt to define it in terms of the use of LISP!). What follows is a discussion of the problem that I previously posted in 1985 (I've made a few minor changes this time)! -- Some inadequate definitions of AI ------------------------------ Marvin Minsky once defined Artificial Intelligence as '... the science of making machines do things that would require intelligence if done by men'. I don't know if he still likes this definition, but it is often quoted with approval. A slightly different definition, similar in spirit but allowing for shifting standards, is given in the textbook on AI by Elaine Rich (McGraw-Hill 1983): '.. the study of how to make computers do things at which, at the moment, people are better.' There are several problems with these definitions. (a) They suggest that AI is primarily a branch of engineering concerned with making machines do things (though Minsky's use of the word 'science' hints at a study of general principles). (b) Perhaps the main objection is their concern with WHAT is done rather than HOW it is done. There are lots of things computers do that would require intelligence if done by people but which have nothing to do with AI, because there are unintelligent ways of getting them done if you have enough speed. E.g. calculators can do complex sums which would require intelligence if done by people. Even simple sums done by a very young child would be regarded as an indication of high intelligence, though not if done by a simple mechanical calculator. Was building calculators to go faster or be more accurate than people once AI? For Rich, does it matter in what way people are currently better? (c) Much AI (e.g. work reported at IJCAI) is concerned with studying general principles in a way that is neutral as to whether it is used for making new machines or explaining how existing systems (e.g. people or squirrels) work. For instance, John McCarthy is said to have coined the term 'Artificial Intelligence' but it is clear that his work is of this more general kind, as is much of the work by Minsky and many others in the field. Many of those who use computers in AI do so merely in order to test, refine, or demonstrate their theories about how people do something, or, more profoundly, because only with the aid of computational concepts can we hope to express theories with rich enough explanatory power. (Which does not mean that present-day computational concepts are sufficient.) For these reasons, the 'Artificial' part of the name is a misnomer, and 'Cognitive Science' or 'Computational Cognitive Science' or 'Epistemics' might have been better names. But it is too late to change the name now, despite the British Alvey Programme's silly use of "IKBS" (Intelligent Knowledge Based Systems) instead of "AI" -- Towards a better definition of AI ------------------------------ Winston, in the second edition of his book on AI (Addison Wesley, 1984) defines AI as 'the study of ideas that enable computers to be intelligent', but quickly moves on to identify two different goals: 'to make computers more useful' 'to understand the principles that make intelligence possible'. His second goal captures the spirit of my complaint about the other definitions. (I made similar points in my book 'The Computer Revolution in Philosophy' (Harvester Press and Humanities Press, 1978; now out of print)). All this assumes that we know what intelligence is: and indeed we can recognise instances even when we cannot define it, as with many other general concepts, like 'cause' 'mind' 'beauty' 'funniness'. Can we hope to have a study of general principles concerning X without a reasonably clear definition of X? Since almost any behaviour can be the product of either an intelligent system (e.g. using false or incomplete beliefs or bizarre motives), or an unintelligent system (e.g. an enormously fast computer using an enormously large look-up table) it is important to define intelligence in terms of HOW the behaviour is produced. -- Towards a definition of Intelligence --------------------------- Intelligent systems are those which: (A) are capable of using structured symbols (e.g. sentences or states of a network; i.e. not just quantitative measures, like temperature or concentration of blood sugar) in a variety of roles including the representation of facts (beliefs), instructions (motives, desires, intentions, goals), plans, strategies, selection principles, etc. NOTE.1. - The set of structures should not be pre-defined: the system should have the "generative" capability to produce new structures as required. The set of uses to which they can be put should also be open ended. (B) are capable of being productively lazy (i.e. able to use the information expressed in the symbols in order to achieve goals with minimal effort). Although it may not be obvious, various kinds of learning capabilities can be derived from (B) which is why I have not included learning as an explicit part of the definition, as some people would. There are many aspects of (A) and (B) which need to be enlarged and clarified, including the notion of 'effort' and how different sorts can be minimised, relative to the system's current capabilities. For instance, there are situations in which the intelligent (productively lazy) thing to do is develop an unintelligent but fast and reliable way to do something which has to be done often. (E.g. learning multiplication tables.) NOTE.2 on above "NOTE.1". I think it is important for intelligence as we conceive it that the mechanisms used should not have any theoretical upper bound to the complexity of the structures with which they can cope, though they may have practical (contingent) limits such as memory limits, and addressing limits..... (The notion of "generative power", i.e. which of a mechanism's limits are theoretically inherent in its design and and which are practical or contingent on the implementation requires further discussion. One test is whether the mechanism could easily make use of more memory if it were provided. A table-lookup mechanism would not be able to extend the table if given more space.) NOTE.3. No definition of intelligence should be regarded as final. As in all science it is to be expected that further investigation will lead to revision of the basic concepts used to define the field. Starting from a suitable (provisional) notion of what an intelligent system is, I would then define AI as the study of principles relevant to explaining or designing actual and possible intelligent systems, including the investigation of both general design requirements and particular implementation tradeoffs. The reference to 'actual' systems includes the study of human and animal intelligence and its underlying principles, and the reference to 'possible' systems covers principles of engineering design for new intelligent systems, as well as possible organisms that might develop one day. NOTE.4: this definition subsumes connectionist (PDP) approaches to the study of intelligence. There is no real conflict between connectionism and AI as conceived of by their broad minded practitioners. The study of ranges of design possibilities (what the limits and tradeoffs are, how different possibilities are related, how they can be generated, etc.) is a part of any theoretical understanding, and good AI MUST be theoretically based. There is lots of bad AI -- what John McCarthy once referred to as the 'look Ma, no hands' variety. The definition of intelligence could be tied more closely to human and animal intelligence by requiring the ability to cope with multiple motives in real time, with resource constraints, in an environment which is partly friendly partly unfriendly. But probably (B) can be interpreted as including all this as a special case! More generally, it is necessary to say something about the nature of the goals and the structure of the environment in which they are to be achieved. But I have gone on long enough. Conclusion: any short and simple definition of AI is likely to be shallow, one-sided, or just wrong as an description of the range of existing AI work. Aaron Sloman, School of Cognitive and Computing Sciences, Univ of Sussex, Brighton, BN1 9QN, England INTERNET: aarons%uk.ac.sussex.cogs@nsfnet-relay.ac.uk aarons%uk.ac.sussex.cogs%nsfnet-relay.ac.uk@relay.cs.net JANET aarons@cogs.sussex.ac.uk BITNET: aarons%uk.ac.sussex.cogs@uk.ac or aarons%uk.ac.sussex.cogs%ukacrl.bitnet@cunyvm.cuny.edu UUCP: ...mcvax!ukc!cogs!aarons or aarons@cogs.uucp From cs!rutgers!iuvax!cica!ctrsol!IDA.ORG!rwex Fri Aug 18 09:56:13 EDT 1989 Article 4733 of comp.ai: Path: cs!rutgers!iuvax!cica!ctrsol!IDA.ORG!rwex >From: rwex@IDA.ORG (Richard Wexelblat) Newsgroups: comp.ai Subject: Re: Is there a definition of AI? Keywords: defining AI Date: 11 Aug 89 11:36:20 GMT References: <1213@syma.sussex.ac.uk> Reply-To: rwex@csed-42.UUCP (Richard Wexelblat) Organization: IDA, Alexandria, VA Lines: 27 In article <1213@syma.sussex.ac.uk> aarons@syma.sussex.ac.uk (Aaron Sloman) writes: >kim@watsup.waterloo.edu (T. Kim Nguyen) writes: >> Anyone seen any mind-blowing (I mean, *GOOD*) definitions of AI? All >> the books seem to gloss over it... >Most people who attempt to define AI give limited definitions based >on ignorance of the breadth of the field. E.g. people who know >nothing about work on computer vision, speech, or robotics often >define AI as if it were all about expert systems. (I even once >saw an attempt to define it in terms of the use of LISP!). A semi-jocular definition I have often quoted (sorry, I don't know the source, I first saw it in net.jokes) is: AI is making computers work like they do in the movies. Clearly, this is circular and less than helpful operationally. But it's a good way to set the scene, especially with layfolks. A problem with the breadth of AI is that as soon as anything begins to be successful, it's not considered AI anymore--as if the opprobrium of being associated with the AI community were something to get away from as soon as possible. Ask someone in NatLang or Robot Vision if they're doing AI. -- --Dick Wexelblat |I must create a System or be enslav'd by another Man's; | (rwex@ida.org) |I will not Reason and Compare: my business is to Create.| 703 824 5511 | -Blake, Jerusalem | From rutgers!cs.utexas.edu!csd4.milw.wisc.edu!bionet!agate!shelby!lindy!news Fri Aug 18 09:56:30 EDT 1989 Article 4736 of comp.ai: Path: sunybcs!rutgers!cs.utexas.edu!csd4.milw.wisc.edu!bionet!agate!shelby!lindy!news >From: GA.CJJ@forsythe.stanford.edu (Clifford Johnson) Newsgroups: comp.ai Subject: Re: Is there a definition of AI? Date: 11 Aug 89 17:18:04 GMT Sender: news@lindy.Stanford.EDU (News Service) Distribution: usa Lines: 51 Here's a footnote I wrote describing "AI" in a document re nuclear "launch on warning" that only mentioned the term in passing. I'd be interested in criticism. It does seem a rather arbitrary term to me. Coined by John McCarthy at Dartmouth in the 1950s, the phrase "Artificial Intelligence" is longhand for computers. Today's machines think. For centuries, classical logicians have pragmatically defined thought as the processing of raw perceptions, comprising the trinity of: categorization of perceptions (Apprehension); comparison of categories of perceptions (Judgment); and the drawing of inferences from connected comparisons (Reason). AI signifies the performance of these definite functions by computers. AI is also a buzz-term that salesmen have applied to virtually all 1980's software, but which to data processing professionals especially connotes software built from large lists of axiomatic "IF x THEN y" rules of inference. (Of course, all programs have some such rules, and, viewed at the machine level, are logically indistinguishable.) The idiom artificial intelligence is curiously convoluted, being applied more often where the coded rules are rough and heuristic (i.e. guesses) rather than precise and analytic (i.e. scientific). The silly innuendo is that AI codifies intuitive expertise. Contrariwise, most AI techniques amount to little more than brute trial-and-error facilitated by rule-of-thumb short-cuts. An analogy is jig-saw reconstruction, which proceeds by first separating pieces with corners and edges, and then crudely trying to find adjacent pairs by exhaustive color and shape matching trials. This analogy should be extended by adding distortion to all pieces of the jig-saw, so that no fit is perfect, and by repainting some, removing other, and adding a few irrelevant pieces. A most likely, or least unlikely, fit is sought. Neural nets are computers programmed with an algorithm for tailoring their rules of thumb, based on statistical inference from a large number of sample observations for which the correct solution is known. In effect, neural nets induce recurrent patterns from input observations. They are limited in the patterns that they recognize, and are stumped by change. Their programmed rules of thumb are not more profound, although they are more complicated, raw "IF... THEN" constructs. Neural nets derive their conditional branchings from underlying rules of statistical inference, and cannot extrapolate beyond the fixations of their induction algorithm. Like regular AI applications, they must select an optimal hypotheses from a simple, predefined set. Thus, all AI applications are largely probabilistic, as exemplified by medical diagnosis and missile attack warning. In medical diagnosis, failure to use and heed a computer can be grounds for malpractice, yet software bugs have gruesome consequences. Likewise, missile attack warning deters, yet puts us all at risk. From rutgers!tut.cis.ohio-state.edu!ucbvax!decwrl!nsc!voder!berlioz!andrew Fri Aug 18 09:56:49 EDT 1989 Article 4740 of comp.ai: Path: sunybcs!rutgers!tut.cis.ohio-state.edu!ucbvax!decwrl!nsc!voder!berlioz!andrew >From: andrew@berlioz (Lord Snooty @ The Giant Poisoned Electric Head ) Newsgroups: comp.ai Subject: Re: Is there a definition of AI? Summary: arrant rubbish Date: 12 Aug 89 05:03:39 GMT References: <4298@lindy.Stanford.EDU> Distribution: usa Organization: National Semiconductor, Santa Clara Lines: 12 In article <4298@lindy.Stanford.EDU>, GA.CJJ@forsythe.stanford.edu (Clifford Johnson) writes: > [Neural nets] are limited in the patterns that they > recognize, and are stumped by change. * flame bit set * Go read about Adaptive Resonance Theory (ART) before making sweeping and false generalisations of this nature! -- ........................................................................... Andrew Palfreyman There's a good time coming, be it ever so far away, andrew@berlioz.nsc.com That's what I says to myself, says I, time sucks jolly good luck, hooray! From rutgers!tut.cis.ohio-state.edu!ucbvax!agate!shelby!lindy!news Fri Aug 18 09:57:17 EDT 1989 Article 4742 of comp.ai: Path: sunybcs!rutgers!tut.cis.ohio-state.edu!ucbvax!agate!shelby!lindy!news >From: GA.CJJ@forsythe.stanford.edu (Clifford Johnson) Newsgroups: comp.ai Subject: Re: Is there a definition of AI? Date: 12 Aug 89 18:37:38 GMT Sender: news@lindy.Stanford.EDU (News Service) Distribution: usa Lines: 12 In <615@berlioz.nsc.com>, Lord Snooty writes: >In <4298@lindy.Stanford.EDU>, Clifford Johnson writes: >> [Neural nets] are limited in the patterns that they >> recognize, and are stumped by change. >Go read about Adaptive Resonance Theory (ART) before making sweeping >and false generalisations of this nature! I would have thought stochastic convergence theory more relevant than resonance theory. What exactly is your point, and what, specifically, should I read? From rutgers!usc!apple!voder!berlioz!andrew Fri Aug 18 09:57:31 EDT 1989 Article 4745 of comp.ai: Path: sunybcs!rutgers!usc!apple!voder!berlioz!andrew >From: andrew@berlioz (Lord Snooty @ The Giant Poisoned Electric Head ) Newsgroups: comp.ai Subject: Re: Is there a definition of AI? Summary: reference citation Date: 12 Aug 89 20:39:19 GMT References: <4318@lindy.Stanford.EDU> Distribution: usa Organization: National Semiconductor, Santa Clara Lines: 26 In article <4318@lindy.Stanford.EDU>, GA.CJJ@forsythe.stanford.edu (Clifford Johnson) writes: > >In <4298@lindy.Stanford.EDU>, Clifford Johnson writes: > >> [Neural nets] are limited in the patterns that they recognize, > >> and are stumped by change. > *flame bit set* > >Go read about Adaptive Resonance Theory (ART) before making sweeping > >and false generalisations of this nature! > > I would have thought stochastic convergence theory more relevant > than resonance theory. > What exactly is your point, and what, specifically, should I read? I refer to "stumped by change", which admittedly is rather inexact in itself. I am not familiar with "stochastic convergence", although perhaps there is another name for it? A characteristic of ART nets is that they are capable of dealing with realtime input and performing dynamic characterisations. A good start would be "Neural Networks & Natural Intelligence" by Stephen Grossberg (ed), 1988, MIT Press. Enjoy. -- ........................................................................... Andrew Palfreyman There's a good time coming, be it ever so far away, andrew@berlioz.nsc.com That's what I says to myself, says I, time sucks jolly good luck, hooray! From ub!zaphod.mps.ohio-state.edu!samsung!cs.utexas.edu!helios!wfsc4!hmueller Thu Aug 30 12:44:08 EDT 1990 Article 7622 of comp.ai: Path: ub!zaphod.mps.ohio-state.edu!samsung!cs.utexas.edu!helios!wfsc4!hmueller >From: hmueller@wfsc4.tamu.edu (Hal Mueller) Newsgroups: comp.ai Subject: Re: What actually is AI? Message-ID: <7838@helios.TAMU.EDU> Date: 30 Aug 90 16:29:37 GMT References: <90241.112651F0O@psuvm.psu.edu> <1990Aug29.183823.25108@msuinfo.cl.msu.edu> <34175@eerie.acsu.Buffalo.EDU> <25392@boulder.Colorado.EDU> <38294@siemens.siemens.com> Sender: usenet@helios.TAMU.EDU Organization: Dept. of Wildlife and Fisheries Sciences, Texas A&M University Lines: 26 In article <38294@siemens.siemens.com> wood@jfred.siemens.edu (Jim Wood) writes: > Artificial Intelligence is a computer science and engineering > discipline which attempts to model human reasoning methods > computationally. I've spent the last year working with a group that tries to build models of ANIMAL reasoning methods; we use the same techniques that you'd apply to any other AI problem. Everything Jim said in his posting is true in this domain as well. Shifting from human to animal reasoning doesn't make the problem any easier. In fact it's rather annoying to be unable to use introspection as a development aid: I can watch myself solve a problem and try to build into a program the techniques I see myself using, but you can't ask an elk or a mountain lion what's going through its brain. All we can do is watch the behavior of our models and compare it to experimentally observed behavior, using the experience of ethologists to guide us. Watching elk in the mountains is much more pleasant than watching a gripper arm pick up blocks, however. -- Hal Mueller Surf Hormuz. hmueller@cs.tamu.edu n270ca@tamunix.Bitnet From ub!zaphod.mps.ohio-state.edu!usc!rutgers!rochester!heron.cs.rochester.edu!yamauchi Tue Sep 4 12:53:42 EDT 1990 Article 7623 of comp.ai: Path: ub!zaphod.mps.ohio-state.edu!usc!rutgers!rochester!heron.cs.rochester.edu!yamauchi >From: yamauchi@heron.cs.rochester.edu (Brian Yamauchi) Newsgroups: comp.ai Subject: Re: What actually is AI? Message-ID: <1990Aug30.175352.2710@cs.rochester.edu> Date: 30 Aug 90 17:53:52 GMT References: <90241.112651F0O@psuvm.psu.edu> <1990Aug29.183823.25108@msuinfo.cl.msu.edu> <34175@eerie.acsu.Buffalo.EDU> <25392@boulder.Colorado.EDU> <38294@siemens.siemens.com> Sender: news@cs.rochester.edu (Usenet news) Reply-To: yamauchi@heron.cs.rochester.edu (Brian Yamauchi) Organization: University of Rochester Computer Science Department Lines: 30 In article <38294@siemens.siemens.com>, wood@jfred.siemens.edu (Jim Wood) writes: > After being in the field for seven years, this is MY informal > definition of Artificial Intelligence: > > Artificial Intelligence is a computer science and engineering > discipline which attempts to model human reasoning methods > computationally. Actually, this sounds more like the (usual) definition of Cognitive Science (since the emphasis is on modeling human reasoning). No doubt if you query a dozen AI researchers, you will receive a dozen different definitions, but my definition would be: Artificial Intelligence is the study of how to build intelligent systems. The term "intelligent" is both fuzzy and open to debate. The usual definition involves symbolic reasoning, but, in my opinion, a better definition would be the ability to generate complex, goal-oriented behavior in a rich, dynamic environment (and perhaps also the ability to learn from experience and extend system abilities based on this learning). But I'm a robotics researcher, so naturally I'm biased :-). _______________________________________________________________________________ Brian Yamauchi University of Rochester yamauchi@cs.rochester.edu Computer Science Department _______________________________________________________________________________ From ub!zaphod.mps.ohio-state.edu!rpi!dali.cs.montana.edu!milton!forbis Tue Sep 4 12:54:34 EDT 1990 Article 7626 of comp.ai: Path: ub!zaphod.mps.ohio-state.edu!rpi!dali.cs.montana.edu!milton!forbis >From: forbis@milton.u.washington.edu (Gary Forbis) Newsgroups: comp.ai Subject: Re: TM's (Was: Re: Searle and Radical Translation) Message-ID: <6889@milton.u.washington.edu> Date: 30 Aug 90 19:58:06 GMT References: <628@ntpdvp1.UUCP> Organization: University of Washington, Seattle Lines: 43 I've been following this line for some time. Ken Presting made me think of an important difference between formal TMs and the day to day computation machines actually do. In article <628@ntpdvp1.UUCP> kenp@ntpdvp1.UUCP (Ken Presting) writes: >> kohout@cme.nist.gov (Robert Kohout) writes: >... The output of real computers is >dependent on the past sequence of inputs, and this is exactly the >phenomenon which concerns me. ... >One reason that change in output over time is important is simply, learning. >I do not see any hope of defining "learning" in terms of machines which >always produce the same output from a given input. > >>If you are saying that a real machine can accept its inputs in little >>chunks, while a TM requires its input up front I maintain that this adds >>nothing to the computing ability of the machine. Obviously, one could take >>the entire input over the life of a real machine and encode it in some >>fashion that could suffice to be the single, "initial" input of a TM. (I am sorry if any feel I have condenced too much. I am trying to keep this article short and pnews requires and equal or greater amount of new text when compared to old text. This lengthens what would otherwise be a short reply to the context setting quoted material.) There is more to real machines than accepting input and producing output. In many cases there is a causal link between previous output and subsequent input. This is an additional reason that no real machine is equivalent to a single TM whose input stream is predetermined. If "the entire input over the life of a real machine" were encoded "in some fashion that could suffice to be the single, 'initial' input of a TM" it would not represent the causal link and as such would require some oracle to be defined. An example. A normal online application session involves separate create, inquiry, update, and delete functions. Unless the imput oracle knows the results of the create prior to actually doing it it cannot encode input for update which relies upon the output of the inquiry. Now I could chop the input into little chunks for each function but then carry some information as input to subsequent calls that are not normally considered part of the input stream (the part Ken is calling remembered.) --gary forbis@milton.u.washington.edu From ub!zaphod.mps.ohio-state.edu!uwm.edu!psuvax1!psuvm!f0o Tue Sep 4 12:56:06 EDT 1990 Article 7630 of comp.ai: Path: ub!zaphod.mps.ohio-state.edu!uwm.edu!psuvax1!psuvm!f0o >From: F0O@psuvm.psu.edu Newsgroups: comp.ai Subject: Re: What actually is AI? Message-ID: <90243.142616F0O@psuvm.psu.edu> Date: 31 Aug 90 18:26:16 GMT References: <90241.112651F0O@psuvm.psu.edu> <1990Aug29.183823.25108@msuinfo.cl.msu.edu> <34175@eerie.acsu.Buffalo.EDU> <6287@jhunix.HCF.JHU.EDU> Organization: Penn State University Lines: 12 In following the threads of my original posting, it seems that there is not one definition of what AI is. However, what my original question was is, what is it that makes one program an AI one, and another one non-AI? Again, I imagine there is not one magical answer to that, but for instance, I'm finishing up a prolog program that plays unbeatable tictactoe. Of course, this is a very simple game, but would it be considered an AI program? If not, how about a checkers or chess program? And it they would be AI programs, what would make them AI, but tictactoe not-AI? [Tim] From ub!zaphod.mps.ohio-state.edu!wuarchive!cs.utexas.edu!uunet!mcsun!unido!uklirb!powers Thu Sep 6 12:45:53 EDT 1990 Article 7642 of comp.ai: Path: ub!zaphod.mps.ohio-state.edu!wuarchive!cs.utexas.edu!uunet!mcsun!unido!uklirb!powers >From: powers@uklirb.informatik.uni-kl.de (David Powers AG Siekmann) Newsgroups: comp.ai Subject: Re: What actually is AI? Message-ID: <6560@uklirb.informatik.uni-kl.de> Date: 3 Sep 90 12:02:27 GMT References: <90241.112651F0O@psuvm.psu.edu> <90243.142616F0O@psuvm.psu.edu> Organization: University of Kaiserslautern, W-Germany Lines: 94 F0O@psuvm.psu.edu writes: > In following the threads of my original posting, it seems that there >is not one definition of what AI is. However, what my original question >was is, what is it that makes one program an AI one, and another one non-AI? >Again, I imagine there is not one magical answer to that, but for instance, >I'm finishing up a prolog program that plays unbeatable tictactoe. Of >course, this is a very simple game, but would it be considered an AI program? >If not, how about a checkers or chess program? And it they would be AI >programs, what would make them AI, but tictactoe not-AI? We have now seen 2 definitions, I prefer to characterize them so: the engineering perspective: to build systems to do the things we can't build systems to do because they require intelligence the psychological perspective: to build systems to do the things we ourselves can do to help us to understand our intelligence The former was the original aim, the latter came from psychology and is represented in Margaret Boden's book: AI and Natural Man. But it is also a natural extension of our familiar introspection. This has now been distinguished with its own name: Cognitive Science. Note that a corollary of the first definition is that once we can build something, then the task no longer lies within artificial intelligence. AI has lost several subfields on this basis, from pattern recognition to chess playing programs to expert systems. I would say the real ai definition is this: the heuristic perspective: to build systems relying on heuristics (rules of thumb) rather than pure algorithms This excludes noughts and crosses (tic-tac-toe) and chess if the progam is dumb and exhaustive (chess) or pre-analyzed and exhaustive (ttt). Unfortunately it could also include expert systems which I see as a spin-off of ai technology and by no means main stream, but expert systems capture conscious knowledge or at least high level knowledge. The capture of the knowledge is straightforward and intrinsically no different from the introspection involved in writing any program - we think "How would I do it by hand?" Of course knowledge engineering techniques can be applied to any domain, even those hard to introspect, by using the techniques with the experts in the field - e.g. on linguists, for natural language. But this won't in general reveal how we are actually really using language. This brings us back to the cognitive science definition. The definition which guides my own work is: to build systems which are capable of modifying their behaviour dynamically by learning This takes the responsibility of acquiring and inputting the heuristics or knowledge from the programmer or knowledge engineer and gives it to the programmer. Machine Learning is a subfield of AI, but somehow central to its future. Expert Systems are also really only still AI in so far as we use AI (=heuristic+learning) techniques in the acquisition of the knowledge base. But there is also a lot of work to be done in establishing the foundations within which learning is possible. Another definition of AI is: Anything written in LISP or PROLOG. This definition (or either half thereof) is believed by some. It is not so silly as it sounds. E.g, PROLOG does have something of the property of automatically finding a way of satisfying specifications, and logic and induction and theorem proving technology are the underpinings of machine learning research. This technology can now be guided by heuristics, and these heuristics can be learned. It's only beginning, but it's exciting! And, of course, you can still misuse any language! I hope this has stirred the pot a bit. David ------------------------------------------------------------------------ David Powers +49-631-205-3449 (Uni); +49-631-205-3200 (Fax) FB Informatik powers@informatik.uni-kl.de; +49-631-13786 (Prv) Univ Kaiserslautern * COMPULOG - Language and Logic 6750 KAISERSLAUTERN * MARPIA - Parallel Logic Programming WEST GERMANY * STANLIE - Natural Language Learning From ub!zaphod.mps.ohio-state.edu!usc!samsung!munnari.oz.au!metro!grivel!gara!pnettlet Thu Sep 6 12:47:56 EDT 1990 Article 7649 of comp.ai: Path: ub!zaphod.mps.ohio-state.edu!usc!samsung!munnari.oz.au!metro!grivel!gara!pnettlet >From: pnettlet@gara.une.oz.au (Philip Nettleton) Newsgroups: comp.ai Subject: What AI is exactly. Summary: Let's look a what AI really is, not just some airy-fairy notions. Message-ID: <3543@gara.une.oz.au> Date: 6 Sep 90 02:43:59 GMT References: <34175@eerie.acsu.Buffalo.EDU> <25392@boulder.Colorado.EDU> <3797@se-sd.SanDiego.NCR.COM> Organization: University of New England, Armidale, Australia Lines: 116 In article <3797@se-sd.SanDiego.NCR.COM>, jim@se-sd.SanDiego.NCR.COM (Jim Ruehlin, Cognitologist domesticus) writes: > In article <38294@siemens.siemens.com> wood@jfred.siemens.edu (Jim Wood) writes: > > Artificial Intelligence is a computer science and engineering > > discipline which attempts to model human reasoning methods > > computationally. > > > > I think this is a pretty good definition, taken from the engineers point > of view. A psychologist might take a different view of the definition/ > purpose of AI. > > One thing I'd include is that it's a cognitive psychological as well as > computer science and engineering discipline. You have to know something > about how people think in order to model human reasoning methods. I think it is a terribly poor definition, actually, for the following reasons: a) Human Intelligence is NOT the only form of intelligence. This is an extremely one eyed view point. Dolphins are extremely intelligent and the only reason we cannot communicate with them to date is because of the extreme differences in our vocal ranges and auditory senses. There is also a huge cultural gap. What concerns do dolphins have? What form does their communication take? We need to know these BEFORE we can even look at syntax and semantics. Hence their intelligence is very alien to ours. b) People tend to assume that at machine cannot be intelligent. Human Intelligence is well documented, much research has been done into Animal Intelligence, but what of Machine Intelligence? Is there a specific type of intelligence that a machine can have? Is there any need to base this intelligence on Human or Animal Intelligence? Saying that AI is modelling "Human Intelligence" is totally inadequate. It may not even be possible because we have such a limited understanding of the processes involved. Artificial Intelligence means: An intelligent system designed by mankind to run on a man-made artifact, ie, a computer. The term Machine Intelligence is more succinct because it identifies the type of intelligence created. Please no arguments about: What is intelligence? This has been discussed ad nauseum, and obviously, we don't know. However, it must exibit intelligence behaviour. With regards to intelligent human behaviour, we can test this with the Turing Test. As for intelligent animal behaviour, there is no appropriate test. And what is intelligent behaviour for a machine? It could be quite alien in appearance from the other two. Let us produce a general requirement for intelligent behaviour: a) The system MUST be able to learn. This implies that the system MUST have a memory for learning to be maintained. Also learning comes in a number of varieties: i) It MUST be able to learn from its own experiences. These can be broken down into further criteria: 1) Learning through trial and error. 2) Learning through observation. 3) Learning through active deduction (see reasoning). ii) It SHOULD be able to learn by instruction, but this is not necessary. At the very least the system MUST have preprogrammed instincts. This is a boot strap for the developing intelligence. Without a starting point, the system cannot progress. b) The system MUST be autonomous. This can be disected as: i) The system MUST be able to effect its environment based on its own independent conclusions. ii) The system MUST be its own master and therefore doesn't require operator intervention. iii) The system MUST be motivated. It must have needs and requirements that can to be satisfied by its own actions. c) The system MUST be able to reason. That is to say, it must use some form of deductive reasoning, based on known facts and capable of producing insights (deductions) which later become known facts. d) The system MUST be self aware. This is related to autonomy, reasoning and learning, but also embodies the need for external senses. Without external senses there is no way of appreciating the difference between "me" and "outside of me". Sensationations of pain and pleasure can provide motivation. It is clear to see that a human easily satisfies these requirements and so is an intelligent system. A cat also satisfies these requirements. So we now have a common basis for known intelligent behaviour. An intelligent machine would need to satisfy these requirements to be classed as an intelligent system. One last point of clarifaction: The ENVIRONMENT in which the intelligent system operates need not be the physical environment of the world around us. It could be a computer environment. I invite responses from those who would like to clarify any points made here or those who would like to extend or advance further points into a constructive debate. But please, if you are hung up on the divity of the human race or you want to bring the Searle debate into this, do us all a favour and refrain. With Regards, Philip Nettleton, Tutor in Computer Science, Department of Maths, Stats, and Computing, The University of New England, Armidale, New South Wales, 2351, AUSTRALIA. From ub!zaphod.mps.ohio-state.edu!rpi!uupsi!njin!princeton!siemens!jfred!wood Fri Sep 21 12:15:41 EDT 1990 Article 7654 of comp.ai: Path: ub!zaphod.mps.ohio-state.edu!rpi!uupsi!njin!princeton!siemens!jfred!wood >From: wood@jfred.siemens.edu (Jim Wood) Newsgroups: comp.ai Subject: Re: What AI is exactly. Message-ID: <38801@siemens.siemens.com> Date: 6 Sep 90 14:43:37 GMT References: <34175@eerie.acsu.Buffalo.EDU> <25392@boulder.Colorado.EDU> <3797@se-sd.SanDiego.NCR.COM> <3543@gara.une.oz.au> Sender: news@siemens.siemens.com Lines: 70 I originally wrote: >> Artificial Intelligence is a computer science and engineering >> discipline which attempts to model human reasoning methods >> computationally. and pnettlet@gara.une.oz.au (Philip Nettleton) writes [and I edit]: >I think it is a terribly poor definition, actually, for the following >reasons: >a) Human intelligence is NOT the only form of intelligence. This is an > extremely one-eyed viewpoint. Dolphins are extremely intelligent, and > the only reason we cannot communicate with them to date is because of > the extreme differences in our vocal ranges and auditory senses. > There is also a huge cultural gap. What concerns do dolphins have? > What form does their communication take? We need to know these > BEFORE we can even look at syntax and semantics. Hence their > intelligence is very alien to ours. Agreed with (a), but I do not recall having implied human intelligence is the only form of intelligence. However, it is certainly the most interesting to artificial intelligence scientists and engineers. From the practical perspective, it is the only type of intelligence which interests industry, from which the purse flows. My definition involves a model of human REASONING methods. The strongest areas of artificial intelligence, in my opinion, are expert systems (modeling the knowledge of an expert), natural language systems (modeling languages and how humans process them), robotics (modeling human sensory and motor functions), and neural networks (modeling the cognitive processes of the human brain). Each of these involves human reasoning. >b) People tend to assume that a machine cannot be intelligent. Human > intelligence is well documented, and much research has been done into > animal intelligence, but what of machine intelligence? Is there a > specific type of intelligence that a machine can have? Is there any > need to base this intelligence on human or animal intelligence? Your reference to machine intelligence is a good one, but it is a mistake to overshadow human intelligence with it in defining artificial intelligence. A machine is no more than an extension of human computability. There is nothing which a machine does which is not a direct product of the exercise of human intelligence. Consequently, machine intelligence is a subset of human intelligence. >Saying that AI is modeling "human intelligence" is totally inadequate. It >may not even be possible because we have such a limited understanding of >the processes involved. I did not say AI models human intelligence. I was very specific to say that it models human reasoning methods. I also believe our knowledge of human reasoning is limited, but that does not stop AI scientists and engineers from developing theories and applications. >Artificial Intelligence means: > An intelligent system designed by mankind to run on a man-made > artifact, for example, a computer. The term Machine Intelligence > is more succinct because it identifies the type of intelligence > created. Artificial intelligence is not a system, any more than computer science is a system. Intelligent systems are the product of artificial intelligence METHODOLOGIES. For example, an expert system is not "artificial intelligence", rather it is the result of applying artificial intelligence methodologies. -- Jim Wood [wood@cadillac.siemens.com] Siemens Corporate Research, 755 College Road East, Princeton, NJ 08540 (609) 734-3643 From ub!zaphod.mps.ohio-state.edu!sdd.hp.com!ucsd!sdcc6!odin!demers Fri Sep 21 12:16:03 EDT 1990 Article 7655 of comp.ai: Path: ub!zaphod.mps.ohio-state.edu!sdd.hp.com!ucsd!sdcc6!odin!demers >From: demers@odin.ucsd.edu (David E Demers) Newsgroups: comp.ai Subject: Re: What actually is AI? Message-ID: <12563@sdcc6.ucsd.edu> Date: 6 Sep 90 19:20:26 GMT References: <90241.112651F0O@psuvm.psu.edu> <90243.142616F0O@psuvm.psu.edu> <6560@uklirb.informatik.uni-kl.de> Sender: news@sdcc6.ucsd.edu Organization: CSE Dept., U. C. San Diego Lines: 53 Nntp-Posting-Host: odin.ucsd.edu In article <6560@uklirb.informatik.uni-kl.de> powers@uklirb.informatik.uni-kl.de (David Powers AG Siekmann) writes: >F0O@psuvm.psu.edu writes: >> In following the threads of my original posting, it seems that there >>is not one definition of what AI is. However, what my original question >>was is, what is it that makes one program an AI one, and another one non-AI? >>Again, I imagine there is not one magical answer to that, but for instance, >>I'm finishing up a prolog program that plays unbeatable tictactoe. Of >>course, this is a very simple game, but would it be considered an AI program? >>If not, how about a checkers or chess program? And it they would be AI >>programs, what would make them AI, but tictactoe not-AI? >We have now seen 2 definitions, I prefer to characterize them so: >the engineering perspective: > to build systems to do the things we can't build systems to do > because they require intelligence the psychological perspective: > to build systems to do the things we ourselves can do to help > us to understand our intelligence [...] >I would say the real ai definition is this: >the heuristic perspective: > to build systems relying on heuristics (rules of thumb) > rather than pure algorithms [...] >The definition which guides my own work is: > to build systems which are capable of modifying their > behaviour dynamically by learning [...] >Another definition of AI is: > > Anything written in LISP or PROLOG. >I hope this has stirred the pot a bit. I'm still looking for the originator of the definition: "AI is the art of making computers act like the ones in the movies" Dave From ub!zaphod.mps.ohio-state.edu!samsung!munnari.oz.au!metro!grivel!gara!pnettlet Fri Sep 21 12:17:48 EDT 1990 Article 7661 of comp.ai: Path: ub!zaphod.mps.ohio-state.edu!samsung!munnari.oz.au!metro!grivel!gara!pnettlet >From: pnettlet@gara.une.oz.au (Philip Nettleton) Newsgroups: comp.ai Subject: What AI is exactly - A follow up. Keywords: intelligence Message-ID: <3569@gara.une.oz.au> Date: 7 Sep 90 04:04:43 GMT Organization: University of New England, Armidale, Australia Lines: 72 lynch@aristotle.ils.nwu.edu (Richard Lynch) writes: > > "No man is an island unto himself." > > Certainly an intelligent machine should be able to handle many things for > itself, but clearly at some point it must be capable of depending on others, > dealing and negotiating with others. > > "TANSTAAFL" -> "There Ain't No Such Thing As A Free Lunch", Cheers! Agreed -> see new definition. forbis@milton.u.washington.edu (Gary Forbis) writes: > > Self awareness does not exist in very young children yet their intelligence > seems apparent to me. Defining the limits of "me" is one of the first tasks > an intelligence has to solve; these limits are fuzzy. Agreed -> see new definition. Let us produce a slightly more refined general requirement for intelligent behaviour: a) The system MUST be able to learn. This implies that the system MUST have a memory for learning to be maintained. Also learning comes in a number of varieties: i) It MUST be able to learn from its own experiences. These can be broken down into further criteria: 1) Learning through trial and error. 2) Learning through observation. 3) Learning through active deduction (see reasoning). ii) It SHOULD be able to learn by instruction, but this is not necessary. At the very least the system MUST have preprogrammed instincts. This is a boot strap for the developing intelligence. Without a starting point, the system cannot progress. b) The system MUST be autonomous. That is to say, it MUST be able to do things by itself (however may choose to accept aid). This can be disected as: i) The system MUST be able to effect its environment based on its own independent conclusions. ii) The system MUST be its own master and therefore doesn't require operator intervention. iii) The system MUST be motivated. It must have needs and requirements that can to be satisfied by its own actions. c) The system MUST be able to reason. That is to say, it must use some form of deductive reasoning, based on known facts and capable of producing insights (deductions) which later become known facts. d) The system MUST be able to develop self awareness. This is related to autonomy, reasoning and learning, but also embodies the need for external senses. Without external senses there is no way of appreciating the difference between "me" and "outside of me". Sensationations of pain and pleasure can provide motivation. With Regards, Philip Nettleton, Tutor in Computer Science, Department of Maths, Stats, and Computing, The University of New England, Armidale, New South Wales, 2351, AUSTRALIA. From ub!acsu.buffalo.edu Fri Sep 21 12:20:25 EDT 1990 Article 7675 of comp.ai: Path: ub!acsu.buffalo.edu >From: dmark@acsu.buffalo.edu (David Mark) Newsgroups: comp.ai Subject: Re: What AI is exactly. Message-ID: <35282@eerie.acsu.Buffalo.EDU> Date: 8 Sep 90 16:13:33 GMT References: <3797@se-sd.SanDiego.NCR.COM> <3543@gara.une.oz.au> <3815@se-sd.SanDiego.NCR.COM> Sender: news@acsu.Buffalo.EDU Organization: SUNY Buffalo Lines: 54 Nntp-Posting-Host: autarch.acsu.buffalo.edu In article <3815@se-sd.SanDiego.NCR.COM> jim@se-sd.SanDiego.NCR.COM (Jim Ruehlin, Cognitologist domesticus) writes: [90 lines deleted] > >Umm, a cat can't reason, or learn in any human sense. ... ^^^ Hope you are not offended, Jim, but I think this claim is just plain silly. Cats, and other mammals, and birds, and indeed even many invertebrates, DO learn things! I remember an article in SCIENCE a few years back that showed that the time required for a butterfly to insert its proboscis into the nectaries of a flower decreases with number of trials. That is "learning", isn't it? And it is A type of learning that humans undoubtedly exhibit. Thus the "any" in the above quote seems inappropriate. (Anyone test a human on time needed to, say, thread a needle?) Yet I don't think I would want to claim that butterflies are "intelligent" in a realistic sense. But, by my everyday definition of "intelligence", cats and crows and many other birds and mammals certainly have it. Their "intelligence" does not seem to be as elaborate or as developed as ours. But they do "learn", and "remember" (experiments with food caching and re-finding in birds; I can find references if you want), and "solve problems" (parrot pulling string "foot over beak" to raise food to its perch), and even "form generalizations". For the latter, I was told of an appartment-raised cat whose owner moved to a house with a front door and a back door. Initially, the cat would "ask" to go out one of the doors, and if it was raining, it would retreat and then "ask" at the other door. But within a few days, the cat, when seeing rain at one door, would NOT attempt the other. It seems obvious that the cat had "generalized" that rain out one door meant rain out the other, or had "learned" that the two doors connect to the "same real world." And as for communication, many animal species have fairly elaborate vocal and behavioral methods for "communicating". And the experiments with signing apes, even if interpreted rather enthusiastically by the authors, seem to indicate abilities at fairly complex communication for these creatures. It seems to me that human "intelligence" differs from the "intelligence" of other vertebrates in degree rather than kind. (I agree that the degree is VERY large in most cases.) Is there any "EVIDENCE" that humans have "kinds" of "intelligence" that no other species exhibits even to a primitive degree? (By the usual standards of science, I would guess that solid "evidence" either way would be pretty hard to come by.) And finally, is the domain or goal of "Artificial Intelligence" really "Artificial HUMAN Intelligence" ? Or do folks mostly want to claim that "Artificial Human Intelligence" is redundant, that "intelligence" is a strictly-human trait? And if so, is it strictly-human BY DEFINITION? And if so, what do we want to call the collective set of cognitive abilities to "learn", "communicate", "solve problems", etc., that many "higher" vertebrates seem to possess? David Mark dmark@acsu.buffalo.edu From ub!acsu.buffalo.edu Fri Sep 21 12:28:02 EDT 1990 Article 7729 of comp.ai: Path: ub!acsu.buffalo.edu >From: dmark@acsu.buffalo.edu (David Mark) Newsgroups: comp.ai Subject: Re: What AI is exactly. Message-ID: <36268@eerie.acsu.Buffalo.EDU> Date: 14 Sep 90 22:35:31 GMT References: <3815@se-sd.SanDiego.NCR.COM> <35282@eerie.acsu.Buffalo.EDU> <3851@se-sd.SanDiego.NCR.COM> Sender: news@acsu.Buffalo.EDU Organization: SUNY Buffalo Lines: 27 Nntp-Posting-Host: autarch.acsu.buffalo.edu In article <3851@se-sd.SanDiego.NCR.COM> jim@se-sd.SanDiego.NCR.COM (Jim Ruehlin , Cognitologist domesticus) writes: > >Perhaps the crux of this problem is the definition of "learning" as >a purely behavioural one. IMO, learning is more than just displaying >certain behaviour. > >>Thus the "any" in the above quote seems inappropriate. > >Agreed, if you look merely at the behavioural aspects of learning. Otherwise, >maybe there's little similarities between the exhibited behaviour in humans >and cats. Jim, it is difficult to discuss issues such as these if people are using the key terms to mean sharply different things. Would you please provide us with the definition of "learning" that you are using, either by making up your own or by quoting some source? I presume that we are not disagreeing much about the facts of animal behavior and human behavior, but are disagreeing about what definitions of "intelligence" and "learn" are appropriate. And since "intelligence" is such a slippery one, let's start with "learn" or "learning". In particular, could you detail what the "non-behavioral" aspects of learning are? David Mark dmark@acsu.buffalo.edu From ub!zaphod.mps.ohio-state.edu!swrinde!ucsd!ogicse!plains!person Fri Sep 21 12:29:17 EDT 1990 Article 7739 of comp.ai: Path: ub!zaphod.mps.ohio-state.edu!swrinde!ucsd!ogicse!plains!person >From: person@plains.NoDak.edu (Brett G. Person) Newsgroups: comp.ai Subject: Re: What actually is AI? Message-ID: <5901@plains.NoDak.edu> Date: 15 Sep 90 23:39:21 GMT References: <90241.112651F0O@psuvm.psu.edu> <90243.142616F0O@psuvm.psu.edu> <6560@uklirb.informatik.uni-kl.de> Organization: North Dakota State University, Fargo Lines: 10 I had an instructor tell me once that AI was anything that hadn't already been done in AI. He said that once the AI app. had been written that no one considered it to be AI anymore -- Brett G. Person North Dakota State University uunet!plains!person | person@plains.bitnet | person@plains.nodak.edu From ub!acsu.buffalo.edu Fri Sep 21 12:31:52 EDT 1990 Article 7744 of comp.ai: Path: ub!acsu.buffalo.edu >From: pmm@acsu.buffalo.edu (patrick m mullhaupt) Newsgroups: comp.ai Subject: Re: What AI is exactly. Message-ID: <36424@eerie.acsu.Buffalo.EDU> Date: 17 Sep 90 04:15:27 GMT References: <25392@boulder.Colorado.EDU> <3797@se-sd.SanDiego.NCR.COM> <3543@gara.une.oz.au> Sender: news@acsu.Buffalo.EDU Organization: SUNY Buffalo Lines: 76 Nntp-Posting-Host: autarch.acsu.buffalo.edu >a) The system MUST be able to learn. >b) The system MUST be autonomous. >c) The system MUST be able to reason. >d) The system MUST be self aware. >It is clear to see that a human easily satisfies these requirements and so is >an intelligent system. A cat also satisfies these requirements. So we now have >a common basis for known intelligent behaviour. An intelligent machine would >need to satisfy these requirements to be classed as an intelligent system. > > With Regards, > > Philip Nettleton, > AUSTRALIA. I don't have any problems with these constraints. I do have a question though. Would a group of individuals, say the congress of the USA, qualify as an "intelligent system"? :-) More generally, do you allow collective intelligences? I would guess that you might not, but your definition seems to allow it. G'day, Patrick Mullhaupt Newsgroups: comp.ai Subject: Followup-To: Distribution: world Organization: SUNY Buffalo Keywords: Newsgroups: comp.ai Subject: Re: What AI is exactly. Summary: Expires: References: <25392@boulder.Colorado.EDU> <3797@se-sd.SanDiego.NCR.COM> <3543@gara.une.oz.au> Sender: Followup-To: Distribution: Organization: SUNY Buffalo Keywords: >a) The system MUST be able to learn. >b) The system MUST be autonomous. >c) The system MUST be able to reason. >d) The system MUST be self aware. >It is clear to see that a human easily satisfies these requirements and so is >an intelligent system. A cat also satisfies these requirements. So we now have >a common basis for known intelligent behaviour. An intelligent machine would >need to satisfy these requirements to be classed as an intelligent system. > > With Regards, > > Philip Nettleton, > AUSTRALIA. I don't have any problems with these constraints. I do have a question though. Would a group of individuals, say the congress of the USA, qualify as an "intelligent system"? :-) More generally, do you allow collective intelligences? I would guess that you might not, but your definition seems to allow it. G'day, Patrick Mullhaupt From ub!zaphod.mps.ohio-state.edu!sdd.hp.com!decwrl!hayes.fai.alaska.edu!accuvax.nwu.edu!mmdf Fri Sep 21 12:32:05 EDT 1990 Article 7746 of comp.ai: Path: ub!zaphod.mps.ohio-state.edu!sdd.hp.com!decwrl!hayes.fai.alaska.edu!accuvax.nwu.edu!mmdf >From: lynch@aristotle.ils.nwu.edu (Richard Lynch) Newsgroups: comp.ai Subject: Re: What is AI Exactly? Message-ID: <12236@accuvax.nwu.edu> Date: 17 Sep 90 15:04:26 GMT Sender: mmdf@accuvax.nwu.edu Lines: 15 Philip Mullhaupt asks, "wouldn't the US Congress" meet the following requirements and thus be classified as 'intelligent': a) Able to learn b) Autonomous c) Able to reason d) Self-aware No, Philip, the US Congress clearly does NOT satisfy c) Able to reason. On a more serious note, one could inject e) Be a single organism, but that would rule out all those sci-fi aliens with only a "collective intelligence", and I don't think we want to do that. "TANSTAAFL" Rich lynch@aristotle.ils.nwu.edu From ub!zaphod.mps.ohio-state.edu!samsung!munnari.oz.au!metro!grivel!gara!pnettlet Fri Sep 21 12:32:38 EDT 1990 Article 7749 of comp.ai: Path: ub!zaphod.mps.ohio-state.edu!samsung!munnari.oz.au!metro!grivel!gara!pnettlet >From: pnettlet@gara.une.oz.au (Philip Nettleton) Newsgroups: comp.ai Subject: What AI is Exactly - Another Update. Keywords: intelligence Message-ID: <3734@gara.une.oz.au> Date: 17 Sep 90 22:27:47 GMT Organization: University of New England, Armidale, Australia Lines: 97 Some new people have recently entered this debate so I thought it was time to repost the definition of an "Intelligent System" that we have developed so far. Pinning this debate back to its origins, we would be interested in hearing from anyone with a CONSTRUCTIVE critism of any part of the definition or any additions they feel are necessary. Remember, the underlying assumption is that to be human is not a necessary condition for being intelligent, this point has been flogged to death in recent postings. Let us produce a slightly more refined "general requirements" for the behaviour of an "intelligent system". ---------------------------------------------------------------------- DEFINITION: GENERAL REQUIREMENTS OF AN INTELLIGENT SYSTEM. a) The system MUST be able to learn. This implies that the system MUST have a memory for learning to be maintained. Also learning comes in a number of varieties: i) It MUST be able to learn from its own experiences. These can be broken down into further criteria: 1) Learning through trial and error. 2) Learning through observation. 3) Learning through active deduction (see reasoning). ii) It SHOULD be able to learn by instruction, but this is not necessary. At the very least the system MUST have preprogrammed instincts. This is a boot strap for the developing intelligence. Without a starting point, the system cannot progress. b) The system MUST be autonomous. That is to say, it MUST be able to do things by itself (however may choose to accept aid). This can be disected as: i) The system MUST be able to effect its environment based on its own independent conclusions. ii) The system MUST be its own master and therefore doesn't require operator intervention. iii) The system MUST be motivated. It must have needs and requirements that can to be satisfied by its own actions. c) The system MUST be able to reason. That is to say, it must use some form of deductive reasoning, based on known facts and capable of producing insights (deductions) which later become known facts. d) The system MUST be able to develop self awareness. This is related to autonomy, reasoning and learning, but also embodies the need for external senses. Without external senses there is no way of appreciating the difference between "me" and "outside of me". Sensationations of pain and pleasure can provide motivation. ---------------------------------------------------------------------- DEFINITION OF TERMS. 1) A "system" CAN be comprised of multiple subsystems, each one of these could be a system in its own right (systems theory). 2) The "environment" in which the system exists MUST be external to the system, but that is as far as the definition of the environment goes (it could be computer generated). 3) The terms "learning", "reasoning" and "autonomy" are BEHAVIOURAL characteristics, further supported by our understanding (to date) of how they MIGHT work. 4) The term "self awareness" is based on learning, reasoning and autonomy, and is the state where the system is aware (has knowledge) of its own existence as separate from its environment. 5) "Intelligence" is a BEHAVIOURAL phenomena displayed by intelligent systems. ---------------------------------------------------------------------- NOTE: If you step OUTSIDE the boundaries of the "definition of terms", your comments will simply be ignored, but feel free to add definitions or modify them if it will help clarify the "general requirements for an intelligent system". With Regards, Philip Nettleton, Tutor in Computer Science, Department of Maths, Stats, and Computing, The University of New England, Armidale, New South Wales, 2351, AUSTRALIA.