Definice pojmů
Vysvětlení "chytrých počítačů" v referenčních knihách
1.Itcanstorealargeamountofinformationandknowledge,andcanreason(includingdeductionandinduction),Ageneral-purposehigh-speedparallelprocessingcomputerwithnon-vonNeumann(John.VonNeumann)structurethatcanexchangeinformationandknowledgewithnaturallanguage,text,sound,graphics,imagesandpeoplewithlearningfunctions.Itisanorganiccombinationofmoderncomputingtechnology,communicationtechnology,artificialintelligenceandbionics,andisatoolforknowledgeprocessing.
Vysvětlení "inteligentního počítače" v akademickém prostředí
1.Intelligentcomputerreferstoanewtypeofcomputerthatcansimulate,extend,expand,andhumanintelligence.ItisverydifferentfromthevonNeumann-typecomputerscurrentlyusedbypeopleintermsofarchitecture,workingmethodsandfunctions.
2.Sincethe1980s,Japan,theUnitedStatesandotherdevelopedcountrieshavebeguntodevelopfifth-generationcomputers,alsoknownasintelligentcomputers.IthighlightstheroleofartificialintelligencemethodsandtechnologiesandconsidersconstructioninsystemdesignTheknowledgebasemanagementsystemandthereasoningengineenablethemachineitselftoreasonandjudgebasedonthestoredknowledge.
3.Inordertorealizethesimulation,extension,andexpansionofhumanintelligenceoncomputers,itisnecessarytocompletelyreformitsarchitecture,workingmethods,processingcapabilities,andinterfacemethods,sothatthecomputercanbecalledItisanintelligentcomputer.Wecanregardtheconstructionofsuchanintelligentmachineasthelong-termgoalofartificialintelligenceresearch.Thefifth-generationcomputerdevelopmentplanproposedbyJapanisanimportantsteptowardsthelong-termgoalandiscalledthefirst-generationintelligentcomputer.
Úvod
Inteligentní počítače
Thereisnogenerallyaccepteddefinitionsofar.Oneofthefoundersofcomputingtheory,A.Turing,definedacomputerasadigitalcomputerthatprocessesdiscreteamountsofinformation.Therearediametricallyopposedviewsontheprincipleofwhetherdigitalcomputerscansimulatehumanintelligence.In1937,A.ChurchandTuringindependentlyproposedhypothesesabouttheequivalenceofhumanthinkingabilityandrecursivefunctionability.Thisunprovenhypothesiswaslaterexpressedbysomeartificialintelligencescholarsas:IfaproblemthatcanbesubmittedtoaTuringmachinecannotbesolvedbyaTuringmachine,theproblemcannotbesolvedbyhumanthinking.Thisschoolinheritedthephilosophicaltraditionofrationalismandreductionismbasedonlogicalthinking,andemphasizedthegreatpotentialofdigitalcomputerstosimulatehumanthinking.Otherscholars,suchasH.Dreyfusandotherphilosophers,affirmativelybelievethatdigitalcomputersbasedonTuringmachinescannotsimulatehumanintelligence.Theybelievethatdigitalcomputerscanonlydoformalizedinformationprocessing,andhumanintelligenceactivitiesmaynotbeformalizedorinformationprocessing,andhumanintellectcannotberegardedasgovernedbydiscreteanddefiniterulesthathavenothingtodowiththeenvironmentalsituation.Operation.Inprinciple,thisschooldoesnotdenythepossibilityofusingmaterialsclosetothehumanbraintoformanintelligentmachine,butthisgeneralizedintelligentmachineisdifferentfromadigitalcomputer.Somescholarsbelievethatnomachinecansimulatehumanintelligence,butmorescholarsbelievethatmostoftheactivitiesinthebraincanbeanalyzedbysymbolsandcalculations.Itmustbepointedoutthatpeople'sunderstandingofcomputingisconstantlydeepeningandbroadening.Somescholarsregardallachievablephysicalprocessesascomputationalprocesses.Genescanalsoberegardedasswitches,andtheoperationsofacellcanalsobeexplainedbycalculations,theso-calledmolecularcalculations.Inthissense,intelligentcomputersinabroadsensearealmostthesameasintelligentmachinesorintelligentmachines.
Differentviewsonsmartcomputersalsocomefromdifferentgoalsandstartingpoints.Oneisthescientificresearchpointofview,emphasizingtheunderstandingofthehumanbrain'smechanism,andrequiringcomputerstoshowintelligentbehaviorinaccordancewiththewaythehumanbrainworks.Theotheristheengineeringperspective,emphasizingtheuseofcomputerstosolvepracticalproblemsthatrequirehumanwisdom,regardlessofwhetherthehumanbrainandthecomputerworkinthesameway.Lookingatanintelligentcomputerfromanengineeringpointofview,whatwecareaboutisthefunctionsithasanddoesnotlimitthewayinwhichitcanberealized.Inotherwords,theso-calledintelligentcomputerreferstotheabilityofperception,recognition,reasoning,learning,etc.,capableofprocessingqualitativeandnotcompletelyuncertainknowledge,capableofcommunicatingwithhumansinnaturallanguage,text,andgraphicsandimages,andhavingtheabilitytoperformintheactualenvironment.Adaptablecomputer.Toachievethisgoalrequireslong-termefforts.
Computer
Acomputerthatexhibitsacertainintelligentbehaviorisnotnecessarilyconsideredtobeintelligent.Inordertodeterminewhetherthecomputerisintelligent,Turingoncedesignedatestmethod,thefamousTuringtest.Imaginethatatesterusesacomputerterminaltocontactthetestedpersonandcomputerseparately(thetestercannotdirectlyseethetestedpersonandthecomputer).Iftheinformationthatanswersthetester’squestioncannotcorrectlydistinguishwhetherthetestedpersonisahumanoracomputer,Ifyoutreatacomputerasanadult,youshouldthinkthatthecomputerisintelligent.Thisisabehavioristviewofsmartphones.Takingtheabilitytoanswerquestionsasacriterionofintelligencehascertainlimitations,becausehumanintelligenceinvolvesmanyaspects,andsomeintelligencessuchasthinkinginimagescannotbesaidinwords.Thiskindoftestisalsodifficulttoreflecttheself-learningandadaptiveability.Ontheonehand,peopleseektorealizeintelligencewithmachines,butontheotherhand,theydon’tbelievethattheautomaticopeningandclosingofelectronicdevicescanrealizehumanthinking.Therefore,whenamethodofimplementingintelligentapplicationsiseffective,itisoftenconsideredtobeaknowntechnology,whichisnodifferentfromothercomputerprograms.Peopletendtothinkthataworkingsystemisasystemthatsimulateshumanthinkingbyamachine.Usefulbutnotreallyintelligent.
Infact,intelligentcomputershavebecomeadynamicdevelopmentconcept,andithasalwaysbeenattheforefrontoftheever-advancingcomputertechnology.Theauthoritativescholarofartificialintelligence,M.Minsky,definesthetaskofartificialintelligencetostudyunsolvedcomputerproblems.Thisviewreflectsthedistinguishingcharacteristicsofartificialintelligenceandintelligentmachineresearchfromotherdisciplines.Intelligentapplicationproblemsoftendonothaveadefinitesolutionalgorithmandusethesearchmethod.Oncepeoplehaveenoughknowledgeaboutacertainproblem,thatis,theyhavefoundadeterministicalgorithmthatdoesnotneedtobesearched,andthebehaviorandeffectoftheproblemcanbepredicted.Generally,itisnolongerconsideredanintelligenceproblem.Fromanapplicationpointofview,ifanartificialintelligenceresearchresultismatureandwidelyadopted,andpeoplealreadyunderstanditsoperatingmechanism,theywillnolongerregarditasanintelligenttechnology.Itcanbepredictedthattheknowledgebase,whichiscurrentlyregardedasthemaincomponentofintelligentcomputers,willberegardedasgeneralcomputertechnologylikedatabasesinthenearfuture.Therefore,ratherthanseeinganintelligentcomputerasacompletelydifferentmachinefromatraditionalcomputer,itisbettertoregarditasahightechnologythatdrivesthecontinuousdevelopmentofcomputers.Thispressureforcesresearchersengagedinintelligentcomputerresearchtoconstantlyproposenewconceptsandnewmethods,andconstantlyovercomenewtechnicaldifficulties.
Thepurposeofdevelopingintelligentcomputersisnottoreplacehumanbrainworkwithcomputers,buttogivefullplaytotherespectivestrengthsofhumansandcomputerstoformacomplementaryandcoordinatedhuman-computercooperationenvironment.Smartcomputersthatarenotverysmartcanmakesmartpeoplesmarter.Intheharmoniousenvironmentofhuman-machinecooperation,peoplearemainlyresponsibleforprovidingawiderangeofcommonsenseandengagingincreativework,andmachinesashumanassistantsperformothertasksthatrequireacertainamountofintelligence.Smartmachinesareoftensmarterinsomeaspectsandstupidinothers.Therefore,todesignanintelligentsystemforhuman-computercollaborationwithhighefficiencyandhighIQ,itisnecessarytoreasonablydeterminewhichthingsaredonebypeopleandwhichthingsaredonebymachines,anditmustbeveryfriendly.Man-machinedialogueinterface.
Themainway
Althoughscholarsfromvariouscountrieshavemadelong-termandunremittingeffortstodevelopintelligentmachines,thereisstillanunresolvedwaytomakecomputersintelligentortoshowintelligentbehavior.Theproblem.Insummary,themainapproachesthathavebeenproposedareasfollows,whicharebasedoncognitivepsychology,neurophysiology,humansociology,andbiologicalevolution.①Symbolprocessingandknowledgeprocessing.Itisthemainstreamofartificialintelligencetotreatintelligenceproblemsassymbolprocessingandknowledgeprocessingproblems.ProfessorNewellandSimonputforwardthehypothesisofthephysicalsymbolsystemintheTuringAwardspeechin1975:thephysicalsymbolsystemisasufficientandnecessaryconditionforintelligentbehavior.Thishypothesisputssymbolprocessingtechnologyatakeypositioninthedevelopmentofintelligentmachines.Theresearchonreasoningmechanismandproblemsolvingtechnologyinthe1960smadepeoplerealizethattheabilityofanintelligentsystemmainlyliesintheknowledgecontainedinthesystemratherthanitsreasoningmechanism.ThisistheknowledgeprincipleadvocatedbyProfessorE.Feigenbaum.Accordingtothisprinciple,thekeytoconstructinganintelligentmachinesystemistoestablishaknowledgebasecontainingalargeamountofcommonsenseandspecializedknowledge.Thetechnicaldifficultyliesintheautomaticacquisitionandmaintenanceofknowledge,andknowledgesharing.Thebasisofthisapproachislogicaltheoryandcognitivepsychology.
②Artificialneuralnetwork.Anotherwaytoconstructanintelligentmachineisrootedintheresearchresultsofneurophysiology,whichistousealargenumberofrelativelysimpleprocessingunits(artificialneurons)toformaneuralnetworkcomputerthroughcomplexinterconnections.Thisapproachemphasizeslarge-scaleparallelism,distributedrepresentationandprocessing,nonlineardynamicsystembehavior,systemtrainingandlearning,andanalogprocessing,etc.Althoughtheartificialneuralnetworkmodelproposedandthevariousartificialneuralnetworksystemsthathavebeendevelopedarefarfromtheneuralnetworkstructureofthehumanbrain,thiskindofoverallstatisticalbehaviorreplaceslogicalreasoning,andsampletrainingandlearningreplacecertaintypesofexecution.Thenewideaofalgorithmisanimpactontraditionalrationalismandreductionism.Neuralnetworkcomputershavepotentialfordevelopmentinpatternrecognitionandlow-levelperceptionsimulation,buttheyalsohavecertainlimitations.Ithasacomplementaryrelationshipwithtraditionalsymbolprocessing.Thecombinationofthetwocangiveplaytotheirrespectiveadvantages.
③Thehierarchicalsocialmodelofintelligence.Theintricatehumansocietyiscomposedofmanyindividualsandgroupsofdifferentlevels.Similarly,intelligentbehaviorcanalsobeseenasanumberofparalleloperationsthatinfluenceeachotheratdifferentlevels.Thelowerthelevel,theworseitsintelligence,andtheprocessingatthelowestlevelshouldbenon-intelligentbehavior.Accordingtothiskindofthinking,thekeyistounderstandhowthecombinationofnon-intelligentactivitiescanemergeintelligentbehavior,andthemysteryshouldlieinitsinterconnection.Thisistheso-called"IntellectualSociety"modeladvocatedbyProfessorMinsky.Thisschoolemphasizestheunderstandingofthelevelofintelligenceandtheconnectionbetweenthevariouspartsofthesystem.Itmainlylooksatthinkingandintelligencefromthebehaviorofhumansociety,anditsrealizationfocusesondistributedartificialintelligenceandcomplexgiantsystems.
④Intelligentsystembasedonbiologicalevolution.Humanintelligenceisproducedthroughextremelylongbiologicalevolution,andevolutionisthesourceofintelligence.Iftheimprovementofmachineintelligenceisalsoregardedasanevolutionaryprocess,itsevolutionspeedwillbemuchfasterthantheformationofhumanintelligence.Thekeytobiologicalevolutionistheabilitytoadaptinadynamicenvironment.Basedonthispointofview,ProfessorBrooksproposedanotherwaytodevelopintelligentmachines:tobuildintelligentsystemswithrealperceptionandactioncapabilitiesintherealworld,andgraduallyimprovetheirintelligencefromsimpletocomplex.Thismethodemphasizesadaptivecontrolandadvocatesthatthereisnoneedtorepresentanintelligentsystemwithoutreasoning.
Eachapproachmentionedabovehasitsowntheoreticalbackgroundandapplicationprospects.Giventhatthefunctionofthebrainistheresultofthecollaborationofthousandsofsubsystemswithdifferentspecializedfunctions,andtheresultofmillionsofbiologicalevolutionaryentanglements,theessenceofhumanintelligencecannotbeattributedtoafewlikewavefunctionsorthethreelawsofkinematics.Thebasicprincipleisthatneat,concise,andbeautiful.Itisalsoimpossibleforsmartphonestobemanufacturedinacertainfixedmode.Researchonintelligentcomputersshouldadoptacomprehensiveandintegratedapproach.Basedontheabove-mentionedapproachesandpossiblenewapproaches,complementarytechnologiessuchasqualitativeandquantitative,digitalandanalog,logicandstatistics,electronicandnon-electronic,etc.shouldbeintegrated,especiallyIntegratetheknowledgestoredinthemachinewiththeexperienceknowledgeofpeople,andgiveplaytotheoveralladvantagesandcomprehensiveadvantagesofthesystem.
Applicationprospects
Intelligentcomputertechnologyisstillimmature,andnowitismainlydoingpatternrecognition,knowledgeprocessinganddevelopmentofintelligentapplications.Althoughtheresultsachievedarestillfarfrompeople'sexpectations,theyhavealreadyproducedobviouseconomicandsocialbenefits.Expertsystemshavebeenwidelyusedinmanagementandscheduling,auxiliarydecision-making,faultdiagnosis,productdesign,educationandconsulting.Significantprogresshasalsobeenmadeintherecognitionandunderstandingoftext,speech,graphicsandimages,aswellasmachinetranslation.Primaryproductsinthisareahavealreadybeenintroducedtothemarket.Theintelligenceofcomputerproductsandtheresearchanddevelopmentofintelligentcomputersystemswillhaveaprofoundimpactonnationaldefense,economy,education,andculture.Computerintelligenceisanimportantdevelopmentdirectionoftheinformationindustryinthe21stcentury.Thedevelopmentofintelligentcomputerswillacceleratethenewindustrialrevolutionmarkedbytheinformationindustry.Theapplicationofsmartcomputerswillamplifyhumanintelligenceandreducetheuseofnaturalresources.Itrequiresverylittleenergyandmaterials,anditsvalueliesmainlyinknowledge.Ontheotherhand,thedevelopmentofintelligentcomputerscanhelppeopleunderstandhumanintelligencemoredeeply,andultimatelyrevealthenatureandmysteriesofintelligence.