artificial intelligence

Detaileddefinitions

Thedefinitionofartificialintelligencecanbedividedintotwoparts,namely"artificial"and"intelligence"."Manual"iseasiertounderstand,andlesscontroversial.Sometimeswehavetoconsiderwhatismanufacturedbyhumans,orwhetherthehuman'sownintelligenceishighenoughtocreateartificialintelligence,andsoon.Butingeneral,the"artificialsystem"isanartificialsystemintheusualsense.

Therearemanyquestionsaboutwhatis"smart".Thisinvolvesotherissuessuchasconsciousness(CONSCIOUSNESS),self(SELF),thinking(MIND)(includingunconsciousthinking(UNCONSCIOUS_MIND))andsoon.Theonlyintelligencethatapersonunderstandsistheintelligenceofthepersonhimself,whichisagenerallyagreedview.Butourunderstandingofourownintelligenceisverylimited,andwealsohavelimitedunderstandingofthenecessaryelementsthatconstitutehumanintelligence,soitisdifficulttodefinewhatis"intelligence"madeby"artificial".Therefore,theresearchofartificialintelligenceofteninvolvestheresearchofhumanintelligenceitself.Otherintelligencerelatedtoanimalsorotherman-madesystemsisalsogenerallyconsideredtoberesearchtopicsrelatedtoartificialintelligence.

Artificialintelligencehasreceivedmoreandmoreattentioninthecomputerfield.Itisalsousedinrobots,economicandpoliticaldecision-making,controlsystems,andsimulationsystems.

ProfessorNelsongaveadefinitionofartificialintelligence:"Artificialintelligenceisadisciplineaboutknowledge-thescienceofhowtoexpressknowledgeandhowtoobtainanduseknowledge."AndanotherMassachusetts,USAProfessorWinstonofthePolytechnicInstitutebelieves:"Artificialintelligenceisthestudyofhowtomakecomputersdointelligentworkthatonlyhumanscandointhepast."Thesestatementsreflectthebasicideasandbasiccontentofartificialintelligence.Thatistosay,artificialintelligenceisthestudyofthelawsofhumanintelligenceactivities,theconstructionofartificialsystemswithcertainintelligence,thestudyofhowtomakecomputerscompletetasksthatrequiredhumanintelligenceinthepast,thatis,thestudyofhowtousecomputersoftwareandhardwaretosimulatecertainhumanintelligenceThebasictheories,methodsandtechniquesofbehavior.

Artificialintelligenceisabranchofcomputerscience.Sincethe1970s,ithasbeenknownasoneoftheworld'sthreecutting-edgetechnologies(spacetechnology,energytechnology,artificialintelligence).Itisalsoconsideredtobeoneofthethreecutting-edgetechnologies(geneticengineering,nanoscience,artificialintelligence)inthe21stcentury.Thisisbecauseithasachievedrapiddevelopmentinthepast30years,hasbeenwidelyusedinmanydisciplines,andhasachievedfruitfulresults.Artificialintelligencehasgraduallybecomeanindependentbranch,bothintheoryandpractice.Intoasystem.

Artificialintelligenceisadisciplinethatstudiestheuseofcomputerstosimulatecertainhumanthinkingprocessesandintelligentbehaviors(suchaslearning,reasoning,thinking,planning,etc.).Itmainlyincludestheprinciplesofcomputerrealizationofintelligenceandmanufacturingsimilartohumans.Brain-smartcomputersenablecomputerstoachievehigher-levelapplications.Artificialintelligencewillinvolvedisciplinessuchascomputerscience,psychology,philosophy,andlinguistics.Itcanbesaidthatalmostalldisciplinesofnaturalsciencesandsocialscienceshaveascopefarbeyondthescopeofcomputerscience.Therelationshipbetweenartificialintelligenceandthinkingscienceistherelationshipbetweenpracticeandtheory,andartificialintelligenceisatthetechnologicalapplicationlevelofthinkingscience.Itisanapplicationbranchofit.Fromtheperspectiveofthinking,artificialintelligenceisnotlimitedtologicalthinking.Itisnecessarytoconsiderimagethinkingandinspirationalthinkingtopromotethebreakthroughdevelopmentofartificialintelligence.Mathematicsisoftenregardedasthebasicscienceofmanydisciplines.Mathematicsalsoentersthefieldoflanguageandthinking.Intelligencedisciplinesmustalsoborrowmathematicaltools.Mathematicsnotonlyplaysaroleinstandardlogic,fuzzymathematics,etc.,butwhenmathematicsentersartificialintelligencedisciplines,theywillpromoteeachotheranddevelopfaster.

Researchvalue

Forexample,heavyscientificandengineeringcalculationswereoriginallybornebythehumanbrain.Nowadays,computerscannotonlyperformsuchcalculations,butalsoperformfasterthanthehumanbrain.,Moreaccurate,socontemporarypeoplenolongerregardthiskindofcalculationasa"complextaskthatrequireshumanintelligencetocomplete".Itcanbeseenthatthedefinitionofcomplexworkchangeswiththedevelopmentofthetimesandtechnologicalprogress.Thespecificgoalsofthesciencewillnaturallydevelopwiththechangesofthetimes.Ontheonehand,itcontinuestomakenewprogress,andontheotherhand,itshiftstomoremeaningfulanddifficultgoals.

Usually,themathematicalfoundationof"machinelearning"is"statistics","informationtheory"and"cybernetics".Italsoincludesothernon-mathematicssubjects.Thistypeof"machinelearning"ishighlydependenton"experience".Computersneedtocontinuouslyacquireknowledgefromtheexperienceofsolvingatypeofproblem,learnstrategies,anduseexperienceknowledgetosolveproblemsandaccumulatenewexperiencewhenencounteringsimilarproblems,justlikeordinarypeople.Wecancallthislearningmethod"continuouslearning".Butinadditiontolearningfromexperience,humanbeingscanalsocreate,thatis,"jumpinglearning."Thisiscalled"inspiration"or"insight"insomesituations.Allalong,thehardestthingtolearnaboutacomputeris"insight."Ormorestrictlyspeaking,itisdifficultforcomputerstolearn"qualitativechangesthatdonotdependonquantitativechanges"inlearningand"practice",anditisdifficulttogodirectlyfromone"quality"toanother"quality",ordirectlyfroma"concept".Toanother"concept".Becauseofthis,the"practice"hereisnotthesameashumanpractice.Theprocessofhumanpracticeincludesbothexperienceandcreation.

Thisisthedreamofintelligentresearchers.

In2013,S.CWang,adataresearcherattheGeneralDataCenterofDijinData,developedanewdataanalysismethod,whichledtoanewmethodforstudyingthepropertiesoffunctions.TheauthorfoundthatthenewdataanalysismethodprovidesawayfortheComputerSocietyto"create".Inessence,thismethodprovidesafairlyeffectivewaytomodelhuman"creativity".Thisapproachisendowedbymathematics,andisa"ability"thatordinarypeoplecannotpossessbutcomputerscanpossess.Sincethen,computersarenotonlygoodatcalculating,butalsogoodatcreatingbecausetheyaregoodatcalculating.Computerscientistsshouldcategoricallydeprivethe"creative"computeroftheoverlycomprehensiveoperatingcapabilities,otherwisethecomputerwilloneday"anti-capture"humanbeings.

Whenlookingbackatthedeductionprocessandmathematicsofthenewmethod,theauthorexpandedhisunderstandingofthinkingandmathematics.Themathematicsisconcise,clear,reliable,andwell-modeled.Inthehistoryofmathematics,thecreativityofmathematicsmastersshineseverywhere.Thesecreativityarepresentedintheformofvariousmathematicaltheoremsorconclusions,andthebiggestfeatureofmathematicaltheoremsisthattheyarebasedonsomebasicconceptsandaxiomsandexpresslogicalstructurescontainingrichinformationinamodularlanguage.Itshouldbesaidthatmathematicsisthesimplestandmoststraightforwardsubjectthatreflects(atleastonetypeof)creativitymodel.

Developmentstage

Inthesummerof1956,agroupofvisionaryyoungscientists,ledbyMcCarthy,Minsky,Rochester,andShennon,gatheredtogetherforjointresearchAnddiscussingaseriesofrelatedissuesaboutusingmachinestosimulateintelligence,andproposedtheterm"artificialintelligence"forthefirsttime,whichmarkedtheofficialbirthoftheemergingdisciplineof"artificialintelligence".IBM's"DeepBlue"computerdefeatedtheworldchesschampionofmankind,anditisaperfectperformanceofartificialintelligencetechnology.

Countingfromtheformalproposalofartificialintelligencein1956,over50years,ithasmadeconsiderableprogressandhasbecomeabroadcross-cuttingandcutting-edgescience.Ingeneral,thepurposeofartificialintelligenceistomakecomputers,machines,thinklikehumans.Ifyouwanttobuildamachinethatcanthink,youmustknowwhatthinkingis,andfurthermore,whatiswisdom.Whatkindofmachineissmart?Scientistshavemadecars,trains,airplanes,radios,etc.,whichimitatethefunctionsofourbodyorgans,butcantheyimitatethefunctionsofthehumanbrain?Sofar,weonlyknowthatthethinginsideourheavenlyspiritcoverisanorganmadeupofbillionsofnervecells.Weknowverylittleaboutthisthing,andimitatingitmaybethemostdifficultthingintheworld.

Whencomputersappeared,humansbegantohaveatoolthatcouldsimulatehumanthinking.Intheyearstocome,countlessscientistsworkedhardforthisgoal.Nowadays,artificialintelligenceisnolongerthepatentofafewscientists.Therearepeopleinthecomputerdepartmentsofalmostalluniversitiesintheworldwhoarestudyingthissubject.Collegestudentsstudyingcomputermustalsolearnsuchacourse.Witheveryone'sunremittingefforts,nowThecomputerseemstohavebecomeverysmart.Forexample,inMay1997,theDEEPBLUEcomputerdevelopedbyIBMdefeatedthechessmasterKasparov(KASPAROV).Youmaynothavenoticedthatinsomeplaces,computershelppeopletoperformothertasksthatwereoriginallyonlyforhumans.Computersplaytheirroleforhumanswithitshighspeedandaccuracy.Artificialintelligencehasalwaysbeenafrontiersubjectofcomputerscience.Computerprogramminglanguages​​andothercomputersoftwarehaveexistedbecauseoftheadvancementofartificialintelligence.

OnMarch4,2019,theSecondSessionoftheThirteenthNationalPeople'sCongressheldapressconference.Thespokespersonoftheconference,ZhangYesui,saidthatlegislationprojectscloselyrelatedtoartificialintelligencehavebeenincludedinthelegislativeplan.

Scientificintroduction

Practicalapplication

Machinevision,fingerprintrecognition,facerecognition,retinarecognition,irisrecognition,palmprintRecognition,expertsystem,automaticplanning,intelligentsearch,theoremproof,game,automaticprogramming,intelligentcontrol,robotics,languageandimageunderstanding,geneticprogramming,etc.

Disciplinarycategory

Artificialintelligenceisafringesubject,whichbelongstotheintersectionofnaturalscienceandsocialscience.

Involvingdisciplines

Philosophyandcognitivesciences,mathematics,neurophysiology,psychology,computerscience,informationtheory,cybernetics,uncertaintytheory

p>

Researchcategory

Naturallanguageprocessing,knowledgerepresentation,intelligentsearch,reasoning,planning,machinelearning,knowledgeacquisition,combinedschedulingproblem,perceptionproblem,patternrecognition,Logicprogramming,softcomputing,inaccurateanduncertainmanagement,artificiallife,neuralnetworks,complexsystems,geneticalgorithms

consciousnessandartificialintelligence

Artificialintelligenceis,initsessence,asimulationoftheinformationprocessofhumanthinking.

Therearetwowaystosimulatehumanthinking.Oneisstructuralsimulation,whichimitatesthestructuralmechanismofthehumanbraintocreatea"brain-like"machine;theotherisfunctionalsimulation,leavingthehumanbraintemporarilyaside.Theinternalstructureissimulatedfromitsfunctionalprocess.Theemergenceofmodernelectroniccomputersisthesimulationofthehumanbrain'sthinkingfunctionandthesimulationoftheinformationprocessofthehumanbrain'sthinking.

Weakartificialintelligencecontinuestodeveloprapidly,especiallyafterthe2008economiccrisis,theUnitedStates,JapanandEuropehopetorealizere-industrializationthroughrobots,etc.Industrialrobotsaredevelopingatafasterratethaneverbefore,andtheyareevenmoredriven.Withcontinuousbreakthroughsinweakartificialintelligenceandrelatedindustries,manytasksthatmustbedonebyhumanscannowbeachievedwithrobots.

Thestrongartificialintelligenceistemporarilyatabottleneckandrequirestheeffortsofscientistsandhumans.

Technicalresearch

Themachineusedtostudythemainmaterialbasisofartificialintelligenceandtheartificialintelligencetechnologyplatformisthecomputer.Thedevelopmenthistoryofartificialintelligenceisthehistoryofthedevelopmentofcomputerscienceandtechnology.Linkedtogether.Inadditiontocomputerscience,artificialintelligencealsoinvolvesmultipledisciplinessuchasinformationtheory,cybernetics,automation,bionics,biology,psychology,mathematicallogic,linguistics,medicine,andphilosophy.Themaincontentofartificialintelligenceresearchincludes:knowledgerepresentation,automaticreasoningandsearchmethods,machinelearningandknowledgeacquisition,knowledgeprocessingsystems,naturallanguageunderstanding,computervision,intelligentrobots,automaticprogramming,etc.

Researchmethods

Nowadays,thereisnounifiedprincipleorparadigmtoguideartificialintelligenceresearch.Therearecontroversiesamongresearchersonmanyissues.Someofthequestionsthathaveremainedinconclusiveforalongtimeare:shouldartificialintelligencebesimulatedfrompsychologicalorneuralaspects?Or,likebirdbiologyistoaviationengineering,humanbiologyhasnothingtodowithartificialintelligenceresearch?Canintelligentbehaviorsbedescribedbysimpleprinciples(suchaslogicoroptimization)?Ormustalargenumberofcompletelyunrelatedproblemsbesolved?

Canintelligencebeexpressedusingadvancedsymbols,suchaswordsandideas?Stillneedtodealwith"sub-symbols"?JOHNHAUGELANDproposedtheconceptofGOFAI(excellentold-fashionedartificialintelligence)andalsoproposedthatartificialintelligenceshouldbeclassifiedasSYNTHETICINTELLIGENCE.[29]Thisconceptwaslateradoptedbysomenon-GOFAIresearchers.

BrainSimulation

Mainarticle:CyberneticsandComputationalNeuroscience

Fromthe1940stothe1950s,manyresearchersExploretheconnectionsbetweenneurology,informationtheoryandcybernetics.Italsocreatedsomepreliminaryintelligenceusingelectronicnetworkstructures,suchasW.GREYWALTER'sTURTLESandJOHNSHOPKINSBEAST.TheseresearchersoftenheldtechnicalassociationmeetingsatPrincetonUniversityandtheRATIOCLUBintheUK.Until1960,mostpeoplehadabandonedthismethod,althoughtheseprincipleswereagainproposedinthe1980s.

Symbolprocessing

Mainarticle:GOFAI

Whenthedigitalcomputerwassuccessfullydevelopedinthe1950s,researchersbegantoexplorehumansCanintelligencebereducedtosymbolprocessing?TheresearchmainlyfocusesonCarnegieMellonUniversity,StanfordUniversityandMassachusettsInstituteofTechnology,andeachhasitsownindependentresearchstyle.JOHNHAUGELANDcallsthesemethodsGOFAI(excellentold-fashionedartificialintelligence).[33]Inthe1960s,thesymbolicmethodmadegreatachievementsinsimulatingadvancedthinkingonsmallproofprograms.Methodsbasedoncyberneticsorneuralnetworksaresecondary.[34]Researchersinthe1960sand1970swereconvincedthatsymbolicmethodscouldeventuallysucceedincreatingmachineswithstrongartificialintelligence,whichwasalsotheirgoal.

CognitivesimulationeconomistsHerbertSimonandAlanNewellstudyhumanproblem-solvingcapabilitiesandtrytoformalizethem.Atthesametime,theylaythefoundationforthebasicprinciplesofartificialintelligence,suchasrecognitionKnowledgescience,operationsresearchandbusinessscience.Theirresearchteamusestheresultsofpsychologicalexperimentstodevelopprogramsthatmimichumanproblem-solvingmethods.ThismethodhasbeeninheritedfromCarnegieMellonUniversityanddevelopedtoitspeakinSOARinthe1980s.Basedonlogic,unlikeAlanNewellandHerbertSimon,JOHNMCCARTHYbelievesthatmachinesdonotneedtosimulatehumanthoughts,butshouldtrytofindabstractreasoningandtheessenceofproblem-solving,regardlessofwhetherpeopleusethesameAlgorithm.HislaboratoryatStanfordUniversityiscommittedtousingformallogictosolveavarietyofproblems,includingknowledgerepresentation,intelligentplanningandmachinelearning.TheUniversityofEdinburghisalsocommittedtologicalmethods,andhascontributedtothedevelopmentofprogramminglanguages​​PROLOGandlogicinotherpartsofEurope.Programmingscience."Anti-logic"StanfordUniversityresearchers(suchasMarvinMinskyandSeymourPipette)foundthattosolvethedifficultproblemsofcomputervisionandnaturallanguageprocessing,specialsolutionsareneeded-theyarguethatthereisnosimpleandGeneralprinciples(suchaslogic)canachieveallintelligentbehaviors.ROGERSCHANKdescribestheir"inverselogic"methodas"SCRUFFY".Commonsenseknowledgebases(suchasDOUGLENAT'sCYC)areexamplesof"SCRUFFY"AIbecausetheyhavetomanuallywriteacomplexconceptatatime.Basedonknowledge,alarge-capacitymemorycomputerappearedinabout1970.Researchersbegantoconstructknowledgeintoapplicationsoftwareinthreeways.This"knowledgerevolution"ledtothedevelopmentandplanningofexpertsystems,whichwasthefirstsuccessfulformofartificialintelligencesoftware.The"knowledgerevolution"alsomadepeoplerealizethatmanysimpleartificialintelligencesoftwaremayrequirealotofknowledge.

Sub-symbolism

Inthe1980s,symbolicartificialintelligencestagnated.Manypeoplebelievedthatsymbolicsystemscouldneverimitateallhumancognitiveprocesses,especiallyItisperception,robotics,machinelearningandpatternrecognition.Manyresearchersbegantopayattentiontothesub-symbolmethodtosolvespecificartificialintelligenceproblems.

Bottom-up,interfaceAGENT,embeddedenvironment(robot),behaviorism,researchersinthefieldofnewAIrobotics,suchasRODNEYBROOKS,negatingsymbolicartificialintelligenceandfocusingonthebasicsofrobotmovementandsurvivalEngineeringissues.Theirworkonceagainfocusedontheviewsofearlycyberneticsresearchers,andatthesametimeproposedtheuseofcontroltheoryinartificialintelligence.Thisisconsistentwiththerepresentationalperceptionargumentinthefieldofcognitivescience:higherintelligencerequiresindividualrepresentations(suchasmovement,perception,andimage).Inthe1980s,DAVIDRUMELHARTandothersonceagainproposedneuralnetworksandconnectionism.Thisandothersub-symbolicmethods,suchasfuzzycontrolandevolutionarycalculation,belongtotheresearchcategoryofcomputationalintelligence.

Statistics

Inthe1990s,artificialintelligenceresearchdevelopedcomplexmathematicaltoolstosolvespecificbranchproblems.Thesetoolsaretrulyscientificmethods,thatis,theresultsofthesemethodsaremeasurableandverifiable,andtheyarealsothereasonforthesuccessofartificialintelligence.Thesharedmathematicallanguagealsoallowscollaborationinexistingdisciplines(suchasmathematics,economics,oroperationsresearch).STUARTJ.RUSSELLandPETERNORVIGpointoutthattheseadvancementsarenolessthanthe"revolution"and"thesuccessofNEATS."Somepeoplecriticizethesetechnologiesforbeingtoofocusedonspecificproblemsandfailingtoconsiderthelong-termgoalofstrongartificialintelligence.

Integrationmethod

SmartAGENTparadigmSmartAGENTisasystemthatperceivestheenvironmentandtakesactionstoachievegoals.Thesimplestintelligentagentsareprogramsthatcansolvespecificproblems.Morecomplexagentsincludehumansandhumanorganizations(suchascompanies).Theseparadigmsallowresearcherstostudyindividualproblemsandfindusefulandverifiablesolutionswithouthavingtoconsiderasinglemethod.Anagentthatsolvesaspecificproblemcanuseanyfeasiblemethod-someagentsusesymbolicandlogicalmethods,andsomearesub-symbolicneuralnetworksorothernewmethods.Theparadigmalsoprovidesresearcherswithacommonlanguagetocommunicatewithotherfields-suchasdecisiontheoryandeconomics(alsousingtheconceptofABSTRACTAGENTS).Theintelligentagentparadigmwaswidelyacceptedinthe1990s.Agentarchitectureandcognitivearchitectureresearchershavedesignedsomesystemstodealwiththeinteractionbetweenintelligentagentsinmulti-ANGENTsystems.Asystemthatcontainssymbolsandsub-symbolsiscalledahybridintelligentsystem,andthestudyofthissystemistheintegrationofartificialintelligencesystems.Thehierarchicalcontrolsystemprovidesabridgebetweenthereaction-levelsub-symbolAIandthehighest-leveltraditionalsymbolAI,whilerelaxingthetimeforplanningandworldmodeling.RODNEYBROOKS'SUBSUMPTIONARCHITECTUREisanearlygradingsystemplan.

Intelligentsimulation

Simulationofmachinevision,hearing,touch,feelingandthinkingmodes:fingerprintrecognition,facerecognition,retinarecognition,irisrecognition,palmprintrecognition,expertsystem,Intelligentsearch,theoremproof,logicalreasoning,game,informationinductionanddialecticalprocessing.

Subjectcategory

Artificialintelligenceisanedgesubject,whichbelongstothethreeinterdisciplinarysubjectsofnaturalscience,socialscience,andtechnicalscience.

Subjectsinvolved

Philosophyandcognitivesciences,mathematics,neurophysiology,psychology,computerscience,informationtheory,cybernetics,uncertainty,bionics,socialstructureandsciencedevelopingview.

Researchcategory

Languagelearningandprocessing,knowledgeexpression,intelligentsearch,reasoning,planning,machinelearning,knowledgeacquisition,combinedschedulingproblem,perceptionproblem,patternrecognition,logicprogramDesign,softcomputing,inaccurateanduncertainmanagement,artificiallife,neuralnetworks,complexsystems,geneticalgorithms,thehumanwayofthinking,themostcriticalproblemistheshapingandimprovementofthemachine'sautonomousandcreativethinkingability.

Securityissues

Artificialintelligenceisstillbeingstudied,butsomescholarsbelievethatitisdangerousforcomputerstohaveIQ,anditmayrebelagainsthumans.Thiskindofhiddendangerhasalsohappenedinmanymovies.Themainkeyistoallowthemachinetohavetheself-consciousnesstoproduceandcontinue.Ifthemachinehastheself-consciousness,itmeansthatthemachinehasthesameorsimilarcreativityashumanbeings.Protectconsciousness,emotionsandspontaneousbehavior.

Implementationmethod

Therearetwodifferentwaystorealizeartificialintelligenceonacomputer.Oneistousetraditionalprogrammingtechniquestomakethesystemappearintelligent,regardlessofwhetherthemethodusedisthesameasthatusedbythehumanoranimalbody.ThismethodiscalledEngineeringMethod(ENGINEERINGAPPROACH),ithasmadeachievementsinsomefields,suchastextrecognition,computerchessandsoon.TheotheristheMODELINGAPPROACH,whichnotonlydependsontheeffect,butalsorequirestheimplementationmethodtobethesameorsimilartothatusedbyhumansorbiologicalorganisms.Geneticalgorithm(GENERICALGORITHM,abbreviatedasGA)andartificialneuralnetwork(ARTIFICIALNEURALNETWORK,abbreviatedasANN)belongtothelattertype.Geneticalgorithmssimulatethegenetic-evolutionmechanismofhumansororganisms,whileartificialneuralnetworkssimulatetheactivityofnervecellsinthehumanoranimalbrain.Inordertogetthesamesmarteffect,bothmethodsareusuallyavailable.Withtheformermethod,itisnecessarytomanuallyspecifytheprogramlogicindetail.Ifthegameissimple,itisstillconvenient.Ifthegameiscomplicatedandthenumberofcharactersandactivityspaceincrease,thecorrespondinglogicwillbeverycomplicated(exponentiallyincrease),andmanualprogrammingwillbeverycumbersomeanderror-prone.Onceanerroroccurs,youmustmodifytheoriginalprogram,recompile,debug,andfinallyprovideuserswithanewversionoranewpatch,whichisverytroublesome.Whenadoptingthelattermethod,theprogrammerhastodesignanintelligentsystem(amodule)foreachroletocontrol.Thisintelligentsystem(module)doesnotunderstandanythingatfirst,justlikeanewbornbaby,butitcanlearnandcanGraduallyadapttotheenvironmentanddealwithvariouscomplexsituations.Thiskindofsystemoftenmakesmistakesatthebeginning,butitcanlearnalesson,anditmaybecorrectedthenexttimeitruns,atleastitwillnotgowrongforever,andthereisnoneedtoreleaseanewversionorpatch.Usingthismethodtorealizeartificialintelligencerequiresprogrammerstohaveabiologicalthinkingmethod,anditismoredifficulttogetstarted.Butonceyouenterthedoor,itcanbewidelyused.Sincethismethoddoesnotrequiredetailedregulationsontherole'sactivityruleswhenprogramming,itisusuallymorelabor-savingthanthepreviousmethodwhenitisappliedtocomplexproblems.

ProfessionalInstitutions

UnitedStates

⒈MASSACHUSETTSINSTITUTEOFTECHNOLOGY

⒉STANFORDUNIVERSITYStanfordUniversity(CA)

⒊CARNEGIEMELLONUNIVERSITY(PA)

⒋UNIVERSITYOFCALIFORNIA-BERKELEY

⒌UNIVERSITYOFWASHINGTON

⒌UNIVERSITYOFWASHINGTON

p>

⒍UNIVERSITYOFTEXAS-AUSTINUniversityofTexasatAustin

⒎UNIVERSITYOFPENNSYLVANIAUniversityofPennsylvania

⒏UNIVERSITYOFILLINOIS-URBANA-CHAMPAIGNUniversityofIllinoisUrbanaThat-Champaign

⒐UNIVERSITYOFMARYLAND-COLLEGEPARK

⒑CORNELLUNIVERSITYCornellUniversity(NY)

⒒UNIVERSITYOFMASSACHUSETTS-AMHERSTUniversityofMassachusettsAMHERSTCampus

⒓GEORGIAINSTITUTEOFTECHNOLOGY

UNIVERSITYOFMICHIGAN-ANNARBORUniversityofMichigan-AnnArbor

⒕UNIVERSITYOFSOUTHERNCALIFORNIAUniversityofSouthernCalifornia

⒖COLUMBIAUNIVERSITYColumbiaUniversity(NY)

UNIVERSITYOFCALIFORNIA-LOSANGELESUniversityofCalifornia,LosAngeles

⒘BROWNUNIVERSITY(RI)

⒙YALEUNIVERSITY(CT)

⒚UNIVERSITYOFCALIFORNIA-SANDIEGOUniversityofCalifornia,SanDiego

⒛UNIVERSITYOFWISCONSIN-MADISONWisconsinUniversityofMadison

China

1,InstituteofAutomation,ChineseAcademyofSciences

2,TsinghuaUniversity

3,PekingUniversity

4,NanjingUniversityofScienceandTechnology

5,BeijingUniversityofScienceandTechnology

6,ChinaScienceandTechnologyUniversity

7,JilinUniversity

8,HarbinInstituteofTechnology

9,BeijingUniversityofPostsandTelecommunications

10,BeijingInstituteofTechnology

p>

11,InstituteofArtificialIntelligence,XiamenUniversity

12,InstituteofIntelligentVehicles,Xi'anJiaotongUniversity

13,InstituteofIntelligentSystemsandIntelligentSoftware,CentralSouthUniversity

14.InstituteofIntelligence,XidianUniversity

15.InstituteofImageandArtificialIntelligence,HuazhongUniversityofScienceandTechnology

16,ChongqingUniversityofPostsandTelecommunications

17,WuhanUniversityofTechnology

MainAchievements

Man-machinegame

February10-17,1996,GARRYKASPAROVdefeated"DeepBlue"4:2(DEEPBLUE).

OnMay3-11,1997,GARRYKASPAROVlosttotheimproved"DeepBlue"2.5:3.5.

InFebruary2003,GARRYKASPAROVdrew3:3with"DEEPJUNIOR".

InNovember2003,GARRYKASPAROVdrew2:2with"X3DGermans"(X3D-FRITZ).

Patternrecognition

Usingthe$patternrecognitionengine,branchinginto2Drecognitionengine,3Drecognitionengine,standingwaverecognitionengineandmulti-dimensionalrecognitionengine

2DrecognitionengineFingerprintrecognition,portraitrecognition,textrecognition,imagerecognition,licenseplaterecognitionhavebeenlaunched;standingwaverecognitionenginehasbeenlaunchedforvoicerecognition;3DrecognitionenginehasbeenlaunchedforfingerprintrecognitionYudaiLinzhongguai(playingsmartversion1.25)

AutomatedEngineering

AutonomousDriving(OSOSystem)

BanknotePrintingFactory(￥AssemblyLine)

FalconSystem(YODDrawing)

KnowledgeEngineering

Takeknowledgeitselfastheprocessingobject,studyhowtouseartificialintelligenceandsoftwaretechnologytodesign,constructandmaintainaknowledgesystem

ExpertSystem

Intelligentsearchengine

Computervisionandimageprocessing

Machinetranslationandnaturallanguageunderstanding

Dataminingandknowledgediscovery

Relatedworks

"VisualArtificialIntelligence":Canmachinesreallythink?Isthehumanmindjustacomplicatedcomputerprogram?Thisbookfocusesononeofthemostdifficultscientificproblemsinhistory,artificialintelligence,andfocusesonsomeofthemaintopicsbehindit.Artificialintelligenceisnotjustafictitiousconcept.Humans’researchonthestructureofintelligentmachinesforhalfacenturyhasshownthatmachinescandefeathumans’greatestchessplayers,andhumanoidrobotscanwalkandinteractwithhumans.Althoughtherehavelongbeendeclarationsthatsmartmachinesarejustaroundthecorner,progressinthisareahasbeenslowanddifficult.Consciousnessandenvironmentaretwomajorproblemsthatplagueresearch.Howshouldwemakesmartmachines?Shoulditfunctionlikeabrain?Doesitrequireabody?FromTuring'sfar-reachinggroundbreakingresearchtotheleapofroboticsandnewartificialintelligence,thisbookclearlypresentsthedevelopmentofartificialintelligenceinthepasthalfcenturytoreaderswithbothpicturesandtexts.

"TheFutureofArtificialIntelligence":interpretstheconnotationofintelligenceandexpoundsthebrainTheworkingprinciple,andtellsushowtomakearealintelligentmachine-suchintelligentmachineswillnolongerjustbeasimpleimitationofthehumanbrain,theirintelligencewillfarexceedthehumanbraininmanyaspects.Hawkinsbelievesthat,fromartificialintelligencetoneuralnetworks,noneoftheearliereffortstoreplicatehumanintelligencehassucceeded.Thereasonisthatpeopledonotreallyunderstandtheconnotationofintelligenceandthehumanbrain.Theso-calledintelligenceistheabilityofthehumanbraintocomparethepastandpredictthefuture.Thebrainisnotacomputer,anditdoesnotfollowthestepsandproduceoutputbasedoninputstepbystep.Thebrainisahugememorysystem.Itstoresexperiencesthatreflectthetruestructureoftheworldtoacertainextent.Itcanrememberthesequenceofeventsandtheirrelationships,andmakepredictionsbasedonmemory.Thefoundationofintelligence,sensation,creativity,andperceptionisthebrain’smemory-predictionsystem...

"ArtificialIntelligencePhilosophy":Theartificialintelligencephilosophyisadevelopmentthataccompaniesmoderninformationtheoryandcomputertechnology.Philosophybranch.Thisbookcollectsfifteenrepresentativepapersbyscholarsinthefieldofartificialintelligenceresearch.Thesepapershavemadegroundbreakingcontributionstothedevelopmentofcomputerscienceandtheestablishmentofartificialintelligencephilosophy.Thesearticlessummarizethehistoryofthedevelopmentofartificialintelligence,thedevelopmenttrendofthesubject,andtheimportanttopicsinartificialintelligence.Amongtheseepoch-makingworksinclude:"ComputerandIntelligence"byAlanTuring,thefatherofmoderncomputertheory;"Mind,Brain,andProgram"byAmericanphilosopherSearle;andJ·E·Hintonetal."DistributedRepresentation"and"EscapefromtheChineseHouse"bytheeditorofthisbookandBritishartificialintelligencescholarM·A·Boden.

"ArtificialIntelligence:AModernMethod":Thisbookcomprehensivelyexpoundsthecorecontentofthefieldofartificialintelligencefromtheperspectiveoftherationalagentwithdetailedandrichinformation,andintroduceseachmajorHisresearchdirectionisararecomprehensivetextbook.Thebookisdividedintoeightparts:thefirstpart"artificialintelligence",thesecondpart"problemsolving",thethirdpart"knowledgeandreasoning",thefourthpart"planning",thefifthpart"uncertainknowledgeandreasoning",thesixthpartPart"Learning",Part7"Communication,PerceptionandAction",Part8"Conclusion".Thisbooknotonlyintroducesalargenumberofbasicconcepts,ideasandalgorithmsindetail,butalsodescribesthemostcutting-edgedevelopmentsinvariousresearchdirections.Atthesametime,itcollectsandorganizesdetailedhistoricaldocumentsandevents.Therefore,thisbookissuitableforresearchersandstudentsofdifferentlevelsandfields.Itcanbeusedasatextbookorteachingguideforundergraduatesandgraduatestudentsintheinformationfieldandrelatedfields,anditcanalsobeusedasareferenceforscientificresearchandengineeringtechniciansinrelatedfields.Book.

Abriefhistoryofdevelopment

ThelegendofartificialintelligencecanbetracedbacktoancientEgypt,butwiththedevelopmentofelectroniccomputerssince1941,technologyhasfinallycreatedmachineintelligence,"artificialintelligence""(ARTIFICIALINTELLIGENCE)wasfirstproposedattheDARTMOUTHSocietyin1956.Sincethen,researchershavedevelopedmanytheoriesandprinciples,andtheconceptofartificialintelligencehasalsoexpanded.Initsshorthistory,Thedevelopmentofartificialintelligenceisslowerthanexpected,butithasbeenmovingforward.Sinceitsappearance40yearsago,manyAIprogramshaveappeared,andtheyhavealsoaffectedthedevelopmentofothertechnologies.

TheComputerAge

Aninventionin1941revolutionizedallaspectsofinformationstorageandprocessing.ThisinventionthatappearedintheUnitedStatesandGermanyatthesametimewastheelectroniccomputer.Acomputerhastooccupyseverallargeair-conditionedrooms,whichisanightmareforprogrammers:thousandsoflinesaresetupjusttorunaprogram.Theimprovedcomputerthatcanstoreprogramsin1949makestheinputprogrambecomeSimpler,andthedevelopmentofcomputertheorygavebirthtocomputerscience,whicheventuallyledtotheemergenceofartificialintelligence.Computers,theinventionofelectronicallyprocessingdata,provideamediumforthepossiblerealizationofartificialintelligence.

AlthoughcomputersprovidethenecessarytechnicalfoundationforAI,itwasnotuntiltheearly1950sthatpeoplenoticedtheconnectionbetweenhumanintelligenceandmachines.NORBERTWIENERwasoneofthefirstAmericanstostudyfeedbacktheory.ThemostfamiliarexampleoffeedbackcontrolItisathermostat.Itcomparesthecollectedroomtemperaturewiththedesiredtemperature,andreactstoturntheheateronorofftocontroltheambienttemperature.Theimportanceofthisresearchonthefeedbackloopliesin:WIENERfromTheoretically,itispointedoutthatallintelligentactivitiesaretheresultoffeedbackmechanism.Thefeedbackmechanismmaybesimulatedbymachines.ThisdiscoveryhasagreatinfluenceonthedevelopmentofearlyAI.

Attheendof1955,NEWWELLandSIMONmadeaprogramcalled"LOGICTHEORIST".ThisprogramisconsideredbymanytobethefirstAIprogram.Itrepresentseachproblemasatreemodel,andthenselectsthemostlikelytogetthecorrectconclusionTosolvetheproblem.Theinfluenceofthe"logicexpert"onthepublicandthefieldofAIresearchmakesitanimportantmilestoneinthedevelopmentofAI.In1956,JOHNMCCARTHY,whoisconsideredthefatherofartificialintelligence,organizedasocietytoManyexpertsandscholarsinterestedinmachineintelligencegatheredforamonthofdiscussion.HeinvitedthemtoVERMONTtoparticipateinthe"DARTMOUTHArtificialIntelligenceSummerResearchConference".Sincethen,thisfieldhasbeennamed"ArtificialIntelligence".AlthoughTheDARTMOUTHSocietywasnotverysuccessful,butitdidgatherthefoundersofAIandlaidthefoundationforfutureAIresearch.

Inthe7yearsaftertheDARTMOUTHconference,AIresearchbegantodeveloprapidly.AlthoughthisThefieldhasnotbeenclearlydefined,andsomeideasintheconferencehavebeenreconsideredandused.CARNEGIEMELLONUniversityandMIThavebeguntoformAIresearchcenters.Researchfacesnewchallenges:thenextstepistoestablishasystemthatcansolveproblemsmoreeffectively,suchasinReducesearchesinthe"LogicExpert";thereisalsotheestablishmentofaself-learningsystem.

In1957,anewprogram,thefirstversionofthe"UniversalProblemSolver"(GPS)wastested.ThisTheprogramwasdevelopedbythesamegroupthatmadethe"LogicExpert".GPSextendedWiener'sfeedbackprincipleandcansolvemanycommonsenseproblems.Twoyearslater,IBMestablishedanAIresearchgroup.HERBERTGELERNETERspent3yearsmakingone.Programstosolvegeometrictheorems.

Asmoreandmoreprogramsemerge,MCCARTHYisbusywithabreakthroughinthehistoryofAI.In1958,MCCARTHYannouncedhisnewachievement:theLISPlanguage.LISPisstilltheretodayUse."LISP"means"listprocessing"(LISTPROCESSING),anditwasquicklyadoptedbymostAIdevelopers.

In1963,MITreceivedaUS$2.2milliongrantfromtheUSgovernment.,Usedtoresearchmachine-aidedrecognition.ThisfundingcomesfromtheAdvancedResearchProjectsAgency(ARPA)oftheDepartmentofDefense,whichhasensuredthattheUnitedStatesisaheadoftheSovietUnionintechnologicalprogress.ThisprogramhasattractedcomputerscientistsfromallovertheworldandacceleratedthedevelopmentofAIresearch.Pace.

Competition

LOEBNER(artificialintelligence)

Usinghumanwisdomtocreateamachinebrain(artificialintelligence)paralleltothehumanbrain,Itisaverytemptingfieldformankind,andmankindhasbeenstrugglingformanyyearstorealizethisdream.Fromtheperspectiveofalanguageresearcher,itisquitedifficulttoallowfreecommunicationbetweenmachinesandhumans,anditmayevenbesaidtobeaquestionthatwillneverbeanswered.Humanlanguageandhumanintelligencearesocomplicatedthatourresearchhasnottouchedtheedgeoftheextendedpartofitsguidingnature.

Alargenumberofprograms

Alargenumberofprogramshaveappearedinthenextfewyears.Oneofthemiscalled"SHRDLU"."SHRDLU"ispartofthe"MicroWorld"project,whichisincludedinthemicroworld(forexample,onlyResearchandprogramminginalimitednumberofgeometricshapes).ResearchersledbyMARVINMINSKYatMITfoundthatinthefaceofsmall-scaleobjects,computerprogramscansolvespatialandlogicalproblems.Otherssuchas"STUDENT"appearedinthelate1960sItcansolvealgebraicproblems."SIR"canunderstandsimpleEnglishsentences.Theresultsoftheseprogramsarehelpfulinprocessinglanguageunderstandingandlogic.

Anotherdevelopmentinthe1970sisexpertsystems.ExpertsystemscanpredictTheprobabilityofacertainsolutionundercertainconditions.Becausethecomputerhadhugecapacityatthattime,itwaspossiblefortheexpertsystemtodrawrulesfromthedata.Themarketapplicationoftheexpertsystemwasverywide.Inthepasttenyears,theexpertsystemwasusedforstockmarketforecaststohelpdoctorsdiagnoseDiseases,aswellasinstructingminerstodeterminethelocationofmineraldeposits.Allofthisispossiblebecauseoftheabilityofexpertsystemstostorerulesandinformation.

ManynewmethodswereusedinAIdevelopmentinthe1970s,suchasMINSKY’sstructuraltheory.Inaddition,DAVIDMARRproposednewtheoriesinmachinevision,forexample,howtodistinguishanimagethroughbasicinformationsuchasshadow,shape,color,borderandtextureofanimage.Byanalyzingthisinformation,youcaninferwhattheimagemaybe.AtthesametimeAnotherachievementisthePROLOGElanguage,whichwasproposedin1972.Duringthe1980s,AIprogressedmorerapidlyandenteredthecommercialfieldmore.In1986,thesalesofAI-relatedsoftwareandhardwareintheUnitedStatesreached425millionU.S.dollars.Expertsystemsareduetotheirutility.Especiallyindemand.CompanieslikeDigitalElectricCompanyusetheXCONexpertsystemtoprogramVAXmainframes.DuPont,GeneralMotorsandBoeingalsorelyheavilyonexpertsystems.Inordertomeettheneedsofcomputerexperts,someproductionexpertsystemsassistintheproductionofsoftwareCompaniessuchasTEKNOWLEDGEandINTELLICORPwereestablished.Inordertofindandcorrecttheerrorsintheexistingexpertsystems,otherexpertsystemshavebeendesigned.

Dailylife

Peoplearebeginningtofeeltheinfluenceofcomputersandartificialintelligencetechnology.Computertechnologyisnolongerjustasmallgroupofresearchersinthelaboratory.Personalcomputersandmanytechnicalmagazineshavebroughtcomputertechnologytothepublic.TherearefoundationsliketheAmericanArtificialIntelligenceAssociation.BecauseoftheneedsofAIdevelopment,therearestillTherewasawaveofresearchersenteringprivatecompanies.Morethan150companieslikeDEC(whichemploysmorethan700employeesinAIresearch)spentatotalofUS$1billiononinternalAIdevelopmentgroups.

OtherAIfieldsalsoenteredthemarketinthe1980s.Oneofthemismachinevision.TheresultsofMINSKYandMARRarenowusedincamerasandcomputersontheproductionlineforqualitycontrol.Althoughstillveryrudimentary,thesesystemshavebeenabletodistinguishtheshapeofobjectsthroughblackandwhitedistinctions.By1985,theUnitedStatesTherearemorethan100companiesproducingmachinevisionsystems,withsalestotalingUS$80million.

Butthe1980swerenotallgoodyearsfortheAI​​industry.ThedevelopmentofAIsystemsin1986-87Demandfell,andtheindustrylostnearlyUS$500million.TwocompanieslikeTEKNOWLEDGEandINTELLICORPlostmorethanUS$6million,accountingforaboutone-thirdofprofits.Thehugelossforcedmanyresearchleaderstocutfunding.AnotherdisappointingItistheso-called"smarttruck"supportedbytheAdvancedResearchProjectsAgencyoftheMinistryofDefense.Thepurposeofthisprojectistodeveloparobotthatcancompletemanybattlefieldtasks.Duetoprojectdefectsandhopelesssuccess,PENTAGONstoppedfundingtheproject.

Artificialintelligencerobots(2photos)

Despitethesesetbacks,AIisstillSlowlyresumedevelopment.NewtechnologieshavebeendevelopedinJapan,suchasfuzzylogicpioneeredintheUnitedStates,whichcanmakedecisionsfromuncertainconditions;andneuralnetworks,whichareregardedaspossiblewaystorealizeartificialintelligence.Inshort,80AIwasintroducedintothemarketinthe1940sandshowedpracticalvalue.Itisbelievedthatitwillbethekeytothe21stcentury.Artificialintelligencetechnologyacceptedandtested.Inthe"DesertStorm"operation,themilitary'ssmartdeviceswithstoodthetestofwar.Manually.Intelligenttechnologyisusedinmissilesystemsandearlywarningdisplaysandotheradvancedweapons.AItechnologyhasalsoenteredthehome.Theincreaseinsmartcomputershasattractedpublicinterest;someapplicationsforAppleandIBMcompatiblecomputers,suchasvoiceandtextrecognition,arealreadyavailableTo;usingfuzzylogic,AItechnologysimplifiescameraequipment.Thegreaterdemandforartificialintelligencerelatedtechnologieshaspromptednewadvancestoemerge.Artificialintelligencehasandwillcontinuetoinevitablychangeourlives.

Comparisonofstrengthsandweaknesses

Amorepopulardefinitionofartificialintelligenceisalsoanearlierdefinitioninthisfield.ItwasdevelopedbyJohnMcCarthyattheDartmouthConferencein1956CONFERENCE):Artificialintelligenceistomakethemachine'sbehaviorlookliketheintelligentbehaviorshownbyhumans.Butthisdefinitionseemstoignorethepossibilityofstrongartificialintelligence(seebelow).Anotherdefinitionreferstoartificialintelligenceastheintelligencedisplayedbyman-mademachines.Generallyspeaking,mostofthedefinitionsofartificialintelligencecanbedividedintofourcategories,namely,machines"thinklikehumans","actlikehumans","thinkrationally"and"actrationally".Here"action"shouldbebroadlyunderstoodastakingaction,ormakingdecisionsforaction,ratherthanphysicalactions.

BOTTOM-UPAI

Thestrongartificialintelligenceviewbelievesthatitispossibletocreaterealreasoning(REASONING)andproblem-solving(PROBLEM_SOLVING).)Intelligentmachines,andsuchmachinescanbeconsideredsentientandself-conscious.Therearetwotypesofstrongartificialintelligence:

Human-likeartificialintelligence,thatis,machinethinkingandreasoningarejustlikehumanthinking.

Non-human-likeartificialintelligence,thatis,machinesproduceperceptionsandconsciousnessthatarecompletelydifferentfromhumans,anduseacompletelydifferentwayofreasoningfromhumans.

TOP-DOWNAI(TOP-DOWNAI)

TheviewofTOP-DOWNAIbelievesthatitisimpossibletoproducerealreasoning(REASONING)andproblem-solving(PROBLEM_SOLVING)intelligentmachines,thesemachinesonlyseemtobeintelligent,buttheydon’treallyhaveintelligence,andtheydon’thaveautonomousconsciousness.

Mainstreamscientificresearchfocusesonweakartificialintelligence,anditisgenerallybelievedthatconsiderableachievementshavebeenmadeinthisresearchfield.Researchonstrongartificialintelligenceisatastandstill.

Philosophicaldebateonstrongartificialintelligence

Theterm"strongartificialintelligence"wasoriginallycoinedbyJohnRogersHillerforcomputersandotherinformationprocessingmachines,Whichisdefinedas:

"Thestrongartificialintelligenceviewbelievesthatcomputersarenotonlyatoolforstudyinghumanthinking;onthecontrary,aslongastheyrunappropriateprograms,thecomputeritselfhasthinking."(JSEARLEINMINDSBRAINSANDPROGRAMS.THEBEHAVIORALANDBRAINSCIENCES,VOL.3,1980)Thisreferstomakingcomputersengageinintelligentactivities.Themeaningofintelligencehereisambiguousanduncertain,asmentionedbelowisanexampleofit.Whenusingacomputertosolveaproblem,youmustknowaclearprocedure.However,therearemanycaseswherepeopletrytosolvetheproblemcleverlyaccordingtotheHEU-RISTICmethodevenwhentheyarenotsureabouttheprocedure.Suchasrecognizingwrittenwords,graphics,sounds,etc.,theso-calledrecognitionmodelisanexample.Inaddition,theimprovementofabilityduetolearning,inductivereasoning,reasoningbasedonanalogy,etc.arealsoexamples.Inaddition,althoughtheprocedureisclear,ittakesalongtimetoimplement.Forsuchaproblem,peoplecanfindaverygoodsolutioninashorttime,suchascompetitivegames.Inaddition,thecomputercannotunderstanditsmeaningwhenitisnotgivensufficientandlogicallycorrectinformation,andpeoplecanalsograspthemeaningbasedonappropriatesupplementaryinformationwhentheyareonlygiveninsufficientandincorrectinformation.Themeaningoflivingit.Naturallanguageisanexample.Processingnaturallanguagewithacomputeriscallednaturallanguageprocessing.

Thedebateonstrongartificialintelligenceisdifferentfromthebroadermonismanddualism(DUALISM)debate.Themainpointoftheargumentis:Iftheonlyworkingprincipleofamachineistotransformencodeddata,doesthismachinehaveathinking?Hillerthinksthisisimpossible.HegaveanexampleofaChineseroomtoillustratethatifthemachineonlyconvertsdata,andthedataitselfisacodedrepresentationofcertainthings,thenyoudon’tunderstandthecorrespondencebetweenthiscodeandtheactualthing.Underthepremiseof,itisimpossibleforthemachinetohaveanyunderstandingofthedataitprocesses.Basedonthisargument,HillerbelievesthatevenifamachinepassestheTuringtest,itdoesnotnecessarilymeanthatthemachineisreallythinkingandconsciouslikeahuman.

Therearealsophilosopherswhoholddifferentviews.InhisbookCONSCIOUSNESSEXPLAINED,DANIELC.DENNETTbelievesthatmanisjustamachinewithasoul.Whydowethinkthathumanscanbeintelligentbutordinarymachinescannot?Hebelievesthatitispossibleforadataconversionmachineliketheabovetohavethinkingandconsciousness.

Somephilosophersbelievethatifweakartificialintelligenceisachievable,thenstrongartificialintelligenceisalsoachievable.Forexample,SIMONBLACKBURNstatedinitsintroductoryphilosophytextbookTHINKthataperson’sactionsthatappeartobe"smart"donotreallymeanthatapersonisreallysmart.Icanneverknowwhetheranotherpersonisreallysmartlikeme,orwhethershe/heisjustassmartasheis.Basedonthisargument,sinceweakartificialintelligencebelievesthatitcanmakeamachinelooksmart,itcannotbecompletelydeniedthatthemachineisreallysmart.BLACKBURNbelievesthatthisisasubjectiveissue.

Itshouldbepointedoutthatweakartificialintelligenceisnotcompletelyopposedtostrongartificialintelligence.Thatistosay,evenifstrongartificialintelligenceispossible,weakartificialintelligenceisstillmeaningful.Atleast,thethingsthatcomputerscandotoday,suchasarithmeticoperations,werethoughttorequireintelligencemorethanahundredyearsago.

PolicyMeasures

OnJune17,2019,theNationalNewGenerationArtificialIntelligenceGovernanceProfessionalCommitteeissuedthe"NewGenerationArtificialIntelligenceGovernancePrinciples-DevelopmentofResponsibleArtificialIntelligence",whichproposedTheframeworkandactionguidelinesforartificialintelligencegovernancehavebeenintroduced.ThisisanimportantachievementforChinatopromotethehealthydevelopmentofanewgenerationofartificialintelligence,strengthenresearchonartificialintelligencelegal,ethical,andsocialissues,andactivelypromotetheglobalgovernanceofartificialintelligence.

Researchtopics

Theresearchdirectionofartificialintelligencehasbeendividedintoseveralsub-fields.Researchershopethatanartificialintelligencesystemshouldhavecertainspecificcapabilities.Thesecapabilitiesarelistedandexplainedbelow.

Solvingproblems

Earlyartificialintelligenceresearchersdirectlyimitatedhumanstoperformstep-by-stepreasoning,justlikethewayhumansthinkwhenplayingaboardgameorperforminglogicalreasoning.Inthe1980sand1990s,usingtheconceptsofprobabilityandeconomics,artificialintelligenceresearchalsodevelopedverysuccessfulmethodstodealwithuncertainorincompleteinformation.

Fordifficultproblems,alargeamountofcomputingresourcesmayberequired,whichmeansthatthereisa"probablecombinationexplosion":whentheproblemexceedsacertainscale,thecomputerwillrequireastronomicalordersofmemoryorcomputingtime.Findingmoreeffectivealgorithmsisapriorityartificialintelligenceresearchproject.

Themodelofhumanproblemsolvingisusuallythequickestandintuitivejudgment,ratherthanconscious,step-by-stepderivation.Earlyartificialintelligenceresearchusuallyusesastep-by-stepderivationmethod.Artificialintelligenceresearchhasmadeprogressinthis"sub-representative"problem-solvingmethod:thematerializedAGENTresearchemphasizestheimportanceofperceivingmovement.Neuralnetworkresearchattemptstoreproducethisskillbysimulatingthebrainstructureofhumansandanimals.

KnowledgeRepresentation

ANONTOLOGYREPRESENTSKNOWLEDGEASASETOFCONCEPTSWITHINADOMAINANDTHERELATIONSHIPSBETWEENTHOSECONCEPTS.

Mainarticle:KnowledgeRepresentationandCommonSenseKnowledgeBase

Planning

Theintelligentagentmustbeabletosetgoalsandachievethesegoals.Theyneedawaytobuildapredictableworldmodel(expressthestateoftheworldinamathematicalmodelandpredicthowtheirbehaviorwillchangetheworld)sothattheycanchoosethemosteffectivebehavior.Intraditionalplanningproblems,theintelligentagentisassumedtobetheonlyinfluentialoneintheworld,sowhatbehavioritwillperformisalreadydetermined.However,ifthisisnotthecase,itmustregularlycheckwhetherthestateoftheworldmodelisconsistentwithitsownpredictions.Ifitdoesnotcomply,itmustchangeitsplan.Therefore,intelligentagentsmusthavetheabilitytoreasonunderuncertainresults.Inmulti-agent,multipleagentsplantoaccomplishcertaingoalsinacooperativeandcompetitivemanner.Usingevolutionaryalgorithmsandgroupwisdomcanachieveanoverallemergentgoal.

Learning

Mainarticle:Machinelearning

Themainpurposeofmachinelearningistoobtainknowledgefromusersandinputdata,etc.,whichcanhelpsolvetheproblem.Multipleproblems,reduceerrors,andimprovetheefficiencyofproblem-solving.Forartificialintelligence,machinelearningisimportantfromtheverybeginning.In1956,attheoriginalDartmouthSummerConference,RaymondSolomonovwroteanarticleaboutprobabilisticmachinelearningwithoutsurveillance:amachineforinductivereasoning.

Naturallanguageprocessing

Mainarticle:Naturallanguageprocessing

Motionandcontrol

Mainarticle:Robotics

Perception

Mainarticle:Machineperception,computervisionandspeechrecognition

Machineperceptionreferstotheabilitytousethedatainputbysensors(suchascameras,microphones,sonarandotherSpecialsensors)theninferthestateoftheworld.Computervisioncananalyzeimageinput.Therearealsovoicerecognition,facerecognitionandobjectrecognition.

Social

Mainarticle:Emotionalcomputing

KISMET,arobotwithsocialabilitiessuchasfacialexpressions

EmotionandsocialskillsSmartAGENTisveryimportant.First,byunderstandingtheirmotivationsandemotionalstates,agentscanpredicttheactionsofothers(thisinvolveselementarygametheory,decisiontheory,andtheabilitytoshapepeople'semotionsandemotionperceptiontesting).Inaddition,forgoodhuman-computerinteraction,intelligentagentsalsoneedtoshowemotions.Atleastitmustappeartodealwithhumanspolitely.Atleast,itshouldhavenormalemotions.

Creativity

Mainarticle:Computercreativity

Asubfieldofartificialintelligence,representingtheory(fromtheperspectiveofphilosophyandpsychology)andpractice(Theoutputofthesystemgeneratedbyaspecificrealizationistheideathatcanbeconsidered,orthesystemrecognizesandevaluatesthecreativity)definedbycreativity.Researchinrelatedfieldsincludesartificialintuitionandartificialimagination.

MultipleIntelligences

Mostresearchershopethattheirresearchwilleventuallybeincorporatedintoamulti-intelligence(calledstrongartificialintelligence)thatcombinesalltheaboveskillsandsurpassesmosthumansAbility.Somepeoplethinkthattoachievetheabovegoals,anthropomorphiccharacteristics,suchasartificialconsciousnessorartificialbrains,mayberequired.Manyoftheaboveproblemsareconsideredtobetheintegrityofartificialintelligence:inordertosolveoneofthem,youmustsolveallofthem.Evenasimpleandspecifictask,suchasmachinetranslation,requiresthemachinetoknowwhatisbeingtalkedabout(knowledge)accordingtotheauthor'sargument(inference),andfaithfullyreproducetheauthor'sintention(emotionalcalculation).Therefore,machinetranslationisconsideredtohaveartificialintelligenceintegrity:itmayrequirestrongartificialintelligence,justlikehumans.

Theinfluenceofartificialintelligence

(1)Theinfluenceofartificialintelligenceonnaturalscience.Insubjectsthatrequiretheuseofmathematicalcomputertoolstosolveproblems,thehelpthatAIbringsisself-evident.Moreimportantly,AIinturnhelpshumanbeingsfinallyunderstandtheformationoftheirownintelligence.

(2)Theimpactofartificialintelligenceontheeconomy.Theexpertsystemgoesdeeperintoallwalksoflifeandbringshugemacrobenefits.AIhasalsopromotedthedevelopmentofthecomputerindustrynetworkindustry.Butatthesametime,italsobringslaboremploymentproblems.DuetotheapplicationofAIintechnologyandengineering,itcanreplacehumansinvarioustechnicalandmentalwork,whichwillcausedrasticchangesinthesocialstructure.

(3)Theimpactofartificialintelligenceonsociety.AIalsoprovidesanewmodelforhumanculturallife.Existinggameswillgraduallydevelopintomoreintelligentinteractiveculturalentertainmentmethods.Today,artificialintelligenceapplicationsingameshavebeendeeplyembeddedinthedevelopmentofmajorgamemanufacturers.

Withthedevelopmentofartificialintelligenceandintelligentrobots,ithastobediscussedthatartificialintelligenceitselfisadvancedresearch,andmodernscientificresearchneedstobecarriedoutwithafuturevision,soitislikelytotouchthebottomlineofethics.Asasensitiveissuethatmaybeinvolvedinscientificresearch,itisnecessarytopreventconflictsthatmayariseassoonaspossible,insteadofwaitingforthecontradictionoftheproblemtobeunresolvablebeforefindingawaytoresolveit.

Applicationfield

Machinetranslation,intelligentcontrol,expertsystem,robotics,languageandimageunderstanding,geneticprogrammingrobotfactory,automaticprogramming,aerospaceapplication,hugeinformationprocessing,Storageandmanagement,executionofcomplexorlarge-scaletasksthatcannotbeperformedbyacompoundedorganism,etc.

Itisworthmentioningthatmachinetranslationisanimportantbranchandthefirstapplicationfieldofartificialintelligence.However,intermsoftheexistingmachinetranslationachievements,thetranslationqualityofthemachinetranslationsystemisstillfarfromtheultimategoal;andthequalityofthemachinetranslationsystemisthekeytothesuccessorfailureofthemachinetranslationsystem.ChinesemathematicianandlinguistProfessorZhouHaizhongoncepointedoutinthepaper"FiftyYearsofMachineTranslation":Toimprovethequalityofmachinetranslation,thefirstthingtodoistosolvetheproblemofthelanguageitselfratherthantheproblemofprogramdesign;onlyrelyonanumberofprogramstodoitThemachinetranslationsystemcertainlycannotimprovethequalityofmachinetranslation;inaddition,itisimpossibleformachinetranslationtoachieve"faithfulness,expressiveness,andelegance"whenhumanshavenotyetunderstoodhowthebrainperformsthefuzzyrecognitionandlogicaljudgmentoflanguage.of.Aftersmarthome,artificialintelligencehasbecomeanewoutletforthehomeapplianceindustry,andChanghongisbecomingthefirsthomeappliancegianttosetoffthiswave.ChanghongreleasedtwonewCHiQsmartTVproducts,featuringmobileremotecontrol,takeawaywatch,watchanytime,andsortedwatchfunctions

Buzzwords

InDecember2017,artificialintelligencewasselectedas"2017"TopTenBuzzwordsinChineseMediaoftheYear".

Reasonforselection:Afteryearsofevolution,thedevelopmentofartificialintelligencehasenteredanewstage.Inordertoseizethemajorstrategicopportunitiesinthedevelopmentofartificialintelligence,buildmycountry'sfirst-moveradvantageinthedevelopmentofartificialintelligence,andacceleratetheconstructionofaninnovativecountryandaworldpowerhouseinscienceandtechnology,onJuly20,2017,theStateCouncilissuedthe"NewGenerationArtificialIntelligenceDevelopmentPlan".The"Plan"putsforwardtheguidingideology,strategicgoals,keytasksandsafeguardmeasuresforthedevelopmentofmycountry'snewgenerationofartificialintelligencein2030,layinganimportantfoundationforthefurtheraccelerateddevelopmentofartificialintelligenceinmycountry.

Developmentstatus

OnJuly13,2021,theInternetSocietyofChinareleasedthe"ChinaInternetDevelopmentReport(2021)".The"Report"showsthatin2020,thescaleoftheartificialintelligenceindustrywillreach303.1billionyuan.

Inthefieldofartificialintelligence,thescaleoftheartificialintelligenceindustrywillmaintainsteadygrowthin2020.Thescaleoftheindustrywillreach303.1billionyuan,ayear-on-yearincreaseof15%,andthegrowthrateisslightlyhigherthantheglobalaverage.TheindustryismainlyconcentratedinBeijing,Shanghai,Guangdong,Zhejiangandotherprovinces.mycountryhasmadesignificantprogressinthefieldofartificialintelligencechips,deeplearningsoftwarearchitecture,andChinesenaturallanguageprocessing.

ThecontentofBaiduBaikeisco-editedbynetizens.Ifyoufindthatthecontentofyourentryisinaccurateorimperfect,pleaseuseyourentryeditingservice(freeofcharge)toparticipateinthecorrection.Gonow>>