8/18/2017 ArtificialintelligenceWikipedia

Artificialintelligence
FromWikipedia,thefreeencyclopedia

Artificialintelligence(AI,alsomachineintelligence,MI)isintelligenceexhibitedbymachines,ratherthan
humansorotheranimals(naturalintelligence,NI).Incomputerscience,thefieldofAIresearchdefinesitself
asthestudyof"intelligentagents":anydevicethatperceivesitsenvironmentandtakesactionsthatmaximize
itschanceofsuccessatsomegoal.[1]Colloquially,theterm"artificialintelligence"isappliedwhenamachine
mimics"cognitive"functionsthathumansassociatewithotherhumanminds,suchas"learning"and"problem
solving".[2]

ThescopeofAIisdisputed:asmachinesbecomeincreasinglycapable,taskconsideredasrequiring
"intelligence"areoftenremovedfromthedefinition,aphenomenonknownastheAIeffect,leadingtothequip
"AIiswhateverhasn'tbeendoneyet."[3]Forinstance,opticalcharacterrecognitionisfrequentlyexcludedfrom
"artificialintelligence",havingbecomearoutinetechnology.[4]CapabilitiesgenerallyclassifiedasAI,asof
2017,includesuccessfullyunderstandinghumanspeech,[5]competingatahighlevelinstrategicgamesystems
(suchaschessandGo[6]),autonomouscars,intelligentroutingincontentdeliverynetworks,military
simulations,andinterpretingcomplexdata.

AIresearchisdividedintosubfields[7]thatfocusonspecificproblems,approaches,theuseofaparticulartool,
ortowardssatisfyingparticularapplications.

Thecentralproblems(orgoals)ofAIresearchincludereasoning,knowledge,planning,learning,natural
languageprocessing(communication),perceptionandtheabilitytomoveandmanipulateobjects.[8]General
intelligenceisamongthefield'slongtermgoals.[9]Approachesincludestatisticalmethods,computational
intelligence,andtraditionalsymbolicAI.ManytoolsareusedinAI,includingversionsofsearchand
mathematicaloptimization,logic,methodsbasedonprobabilityandeconomics.TheAIfielddrawsupon
computerscience,mathematics,psychology,linguistics,philosophy,neuroscience,artificialpsychologyand
manyothers.

Thefieldwasfoundedontheclaimthathumanintelligence"canbesopreciselydescribedthatamachinecan
bemadetosimulateit".[10]Thisraisesphilosophicalargumentsaboutthenatureofthemindandtheethicsof
creatingartificialbeingsendowedwithhumanlikeintelligence,issueswhichhavebeenexploredbymyth,
fictionandphilosophysinceantiquity.[11]SomepeoplealsoconsiderAIadangertohumanityifitprogresses
unabatedly.[12]Attemptstocreateartificialintelligencehaveexperiencedmanysetbacks,includingtheALPAC
reportof1966,theabandonmentofperceptronsin1970,theLighthillReportof1973,thesecondAIwinter
19871993andthecollapseoftheLispmachinemarketin1987.

Inthetwentyfirstcentury,AItechniques,bothhard(usingasymbolicapproach)andsoft(subsymbolic),have
experiencedaresurgencefollowingconcurrentadvancesincomputerpower,sizesoftrainingsets,and
theoreticalunderstanding,andAItechniqueshavebecomeanessentialpartofthetechnologyindustry,helping
tosolvemanychallengingproblemsincomputerscience.[13]RecentadvancementsinAI,andspecificallyin
machinelearning,havecontributedtothegrowthofAutonomousThingssuchasdronesandselfdrivingcars,
becomingthemaindriverofinnovationintheautomotiveindustry.

Contents
1 History
2 Goals
2.1 Reasoning,problemsolving
2.2 Knowledgerepresentation
2.3 Planning
2.4 Learning
2.5 Naturallanguageprocessing

https://en.wikipedia.org/wiki/Artificial_intelligence 1/42
8/18/2017 ArtificialintelligenceWikipedia

2.6 Perception
2.7 Motionandmanipulation
2.8 Socialintelligence
2.9 Creativity
2.10 Generalintelligence
3 Approaches
3.1 Cyberneticsandbrainsimulation
3.2 Symbolic
3.3 Subsymbolic
3.4 Integratingtheapproaches
4 Tools
4.1 Searchandoptimization
4.2 Logic
4.3 Probabilisticmethodsforuncertainreasoning
4.4 Classifiersandstatisticallearningmethods
4.5 Neuralnetworks
4.6 Deepfeedforwardneuralnetworks
4.7 Deeprecurrentneuralnetworks
4.8 Controltheory
4.9 Languages
4.10 Evaluatingprogress
5 Applications
5.1 Competitionsandprizes
5.2 Healthcare
5.3 Automotive
5.4 Finance
5.5 Videogames
6 Platforms
6.1 PartnershiponAI
7 Philosophyandethics
7.1 Thelimitsofartificialgeneralintelligence
7.2 Potentialrisksandmoralreasoning
7.3 Machineconsciousness,sentienceandmind
7.4 Superintelligence
8 Infiction
9 Seealso
10 Notes
11 References
11.1 AItextbooks
11.2 HistoryofAI
11.3 Othersources
12 Furtherreading
13 Externallinks

History
Whilethoughtcapableartificialbeingsappearedasstorytellingdevicesinantiquity,[14]theideaofactually
tryingtobuildamachinetoperformusefulreasoningmayhavebegunwithRamonLlull(c.1300CE).Withhis
Calculusratiocinator,GottfriedLeibnizextendedtheconceptofthecalculatingmachine(WilhelmSchickard
engineeredthefirstonearound1623),intendingtoperformoperationsonconceptsratherthannumbers.[15]
Sincethe19thcentury,artificialbeingsarecommoninfiction,asinMaryShelley'sFrankensteinorKarel
apek'sR.U.R.(Rossum'sUniversalRobots).[16]

https://en.wikipedia.org/wiki/Artificial_intelligence 2/42
8/18/2017 ArtificialintelligenceWikipedia

Thestudyofmechanicalor"formal"reasoningbeganwithphilosophersandmathematiciansinantiquity.The
studyofmathematicallogicleddirectlytoAlanTuring'stheoryofcomputation,whichsuggestedthata
machine,byshufflingsymbolsassimpleas"0"and"1",couldsimulateanyconceivableactofmathematical
deduction.Thisinsight,thatdigitalcomputerscansimulateanyprocessofformalreasoning,isknownasthe
ChurchTuringthesis.[17]Alongwithconcurrentdiscoveriesinneurology,informationtheoryandcybernetics,
thisledresearcherstoconsiderthepossibilityofbuildinganelectronicbrain.[18]Thefirstworkthatisnow
generallyrecognizedasAIwasMcCullouchandPitts'1943formaldesignforTuringcomplete"artificial
neurons".[15]

ThefieldofAIresearchwasbornataworkshopatDartmouthCollegein1956.[19]AttendeesAllenNewell
(CMU),HerbertSimon(CMU),JohnMcCarthy(MIT),MarvinMinsky(MIT)andArthurSamuel(IBM)
becamethefoundersandleadersofAIresearch.[20]Theyandtheirstudentsproducedprogramsthatthepress
describedas"astonishing":[21]computerswerewinningatcheckers,solvingwordproblemsinalgebra,proving
logicaltheoremsandspeakingEnglish.[22]Bythemiddleofthe1960s,researchintheU.S.washeavilyfunded
bytheDepartmentofDefense[23]andlaboratorieshadbeenestablishedaroundtheworld.[24]AI'sfounders
wereoptimisticaboutthefuture:HerbertSimonpredicted,"machineswillbecapable,withintwentyyears,of
doinganyworkamancando".MarvinMinskyagreed,writing,"withinageneration...theproblemofcreating
'artificialintelligence'willsubstantiallybesolved".[25]

Theyfailedtorecognizethedifficultyofsomeoftheremainingtasks.Progressslowedandin1974,inresponse
tothecriticismofSirJamesLighthill[26]andongoingpressurefromtheUSCongresstofundmoreproductive
projects,boththeU.S.andBritishgovernmentscutoffexploratoryresearchinAI.Thenextfewyearswould
laterbecalledan"AIwinter",[27]aperiodwhenobtainingfundingforAIprojectswasdifficult.

Intheearly1980s,AIresearchwasrevivedbythecommercialsuccessofexpertsystems,[28]aformofAI
programthatsimulatedtheknowledgeandanalyticalskillsofhumanexperts.By1985themarketforAIhad
reachedoverabilliondollars.Atthesametime,Japan'sfifthgenerationcomputerprojectinspiredtheU.Sand
Britishgovernmentstorestorefundingforacademicresearch.[29]However,beginningwiththecollapseofthe
LispMachinemarketin1987,AIonceagainfellintodisrepute,andasecond,longerlastinghiatusbegan.[30]

Inthelate1990sandearly21stcentury,AIbegantobeusedforlogistics,datamining,medicaldiagnosisand
otherareas.[13]Thesuccesswasduetoincreasingcomputationalpower(seeMoore'slaw),greateremphasison
solvingspecificproblems,newtiesbetweenAIandotherfieldsandacommitmentbyresearchersto
mathematicalmethodsandscientificstandards.[31]DeepBluebecamethefirstcomputerchessplayingsystem
tobeatareigningworldchesschampion,GarryKasparovon11May1997.[32]

Advancedstatisticaltechniques(looselyknownasdeeplearning),accesstolargeamountsofdataandfaster
computersenabledadvancesinmachinelearningandperception.[33]Bythemid2010s,machinelearning
applicationswereusedthroughouttheworld.[34]InaJeopardy!quizshowexhibitionmatch,IBM'squestion
answeringsystem,Watson,defeatedthetwogreatestJeopardychampions,BradRutterandKenJennings,bya
significantmargin.[35]TheKinect,whichprovidesa3DbodymotioninterfacefortheXbox360andtheXbox
OneusealgorithmsthatemergedfromlengthyAIresearch[36]asdointelligentpersonalassistantsin
smartphones.[37]InMarch2016,AlphaGowon4outof5gamesofGoinamatchwithGochampionLee
Sedol,becomingthefirstcomputerGoplayingsystemtobeataprofessionalGoplayerwithout
handicaps.[6][38]Inthe2017FutureofGoSummit,AlphaGowonathreegamematchwithKeJie,[39]whoat
thetimecontinuouslyheldtheworldNo.1rankingfortwoyears[40][41]

AccordingtoBloomberg'sJackClark,2015wasalandmarkyearforartificialintelligence,withthenumberof
softwareprojectsthatuseAIwithinGoogleincreasedfroma"sporadicusage"in2012tomorethan2,700
projects.Clarkalsopresentsfactualdataindicatingthaterrorratesinimageprocessingtaskshavefallen
significantlysince2011.[42]Heattributesthistoanincreaseinaffordableneuralnetworks,duetoarisein
cloudcomputinginfrastructureandtoanincreaseinresearchtoolsanddatasets.Othercitedexamplesinclude
Microsoft'sdevelopmentofaSkypesystemthatcanautomaticallytranslatefromonelanguagetoanotherand
Facebook'ssystemthatcandescribeimagestoblindpeople.[42]

https://en.wikipedia.org/wiki/Artificial_intelligence 3/42
8/18/2017 ArtificialintelligenceWikipedia

Goals
Theoverallresearchgoalofartificialintelligenceistocreatetechnologythatallowscomputersandmachines
tofunctioninanintelligentmanner.Thegeneralproblemofsimulating(orcreating)intelligencehasbeen
brokendownintosubproblems.Theseconsistofparticulartraitsorcapabilitiesthatresearchersexpectan
intelligentsystemtodisplay.Thetraitsdescribedbelowhavereceivedthemostattention.[8]

ErikSandwellemphasizesplanningandlearningthatisrelevantandapplicabletothegivensituation.[43]

Reasoning,problemsolving

Earlyresearchersdevelopedalgorithmsthatimitatedstepbystepreasoningthathumansusewhentheysolve
puzzlesormakelogicaldeductions.[44]Bythelate1980sand1990s,AIresearchhaddevelopedmethodsfor
dealingwithuncertainorincompleteinformation,employingconceptsfromprobabilityandeconomics.[45]

Fordifficultproblems,algorithmscanrequireenormouscomputationalresourcesmostexperiencea
"combinatorialexplosion":theamountofmemoryorcomputertimerequiredbecomesastronomicalfor
problemsofacertainsize.Thesearchformoreefficientproblemsolvingalgorithmsisahighpriority.[46]

Humanbeingsordinarilyusefast,intuitivejudgmentsratherthanstepbystepdeductionthatearlyAIresearch
wasabletomodel.[47]AIhasprogressedusing"subsymbolic"problemsolving:embodiedagentapproaches
emphasizetheimportanceofsensorimotorskillstohigherreasoningneuralnetresearchattemptstosimulate
thestructuresinsidethebrainthatgiverisetothisskillstatisticalapproachestoAImimicthehumanabilityto
guess.

Knowledgerepresentation

KnowledgerepresentationandknowledgeengineeringarecentraltoAI
research.Manyoftheproblemsmachinesareexpectedtosolvewill
requireextensiveknowledgeabouttheworld.AmongthethingsthatAI
needstorepresentare:objects,properties,categoriesandrelations
betweenobjectssituations,events,statesandtimecausesandeffects
knowledgeaboutknowledge(whatweknowaboutwhatotherpeople
know)andmanyother,lesswellresearcheddomains.Arepresentation
of"whatexists"isanontology:thesetofobjects,relations,concepts,
andpropertiesformallydescribedsothatsoftwareagentscaninterpret
them.Thesemanticsofthesearecapturedasdescriptionlogicconcepts,
roles,andindividuals,andtypicallyimplementedasclasses,properties,
andindividualsintheWebOntologyLanguage.[48]Themostgeneral
ontologiesarecalledupperontologies,whichactasmediatorsbetween Anontologyrepresentsknowledgeasa
setofconceptswithinadomainandthe
domainontologiesthatcoverspecificknowledgeaboutaparticular
relationshipsbetweenthoseconcepts.
knowledgedomain(fieldofinterestorareaofconcern).Suchformal
knowledgerepresentationsaresuitableforcontentbasedindexingand
retrieval,sceneinterpretation,clinicaldecisionsupport,knowledge
discoveryviaautomatedreasoning(inferringnewstatementsbasedonexplicitlystatedknowledge),etc.Video
eventsareoftenrepresentedasSWRLrules,whichcanbeused,amongothers,toautomaticallygenerate
subtitlesforconstrainedvideos.[49].

Amongthemostdifficultproblemsinknowledgerepresentationare:

Defaultreasoningandthequalificationproblem
Manyofthethingspeopleknowtaketheformof"workingassumptions".Forexample,ifabirdcomes
upinconversation,peopletypicallypictureananimalthatisfistsized,sings,andflies.Noneofthese
thingsaretrueaboutallbirds.JohnMcCarthyidentifiedthisproblemin1969[50]asthequalification
problem:foranycommonsenserulethatAIresearcherscaretorepresent,theretendtobeahugenumber
https://en.wikipedia.org/wiki/Artificial_intelligence 4/42
8/18/2017 ArtificialintelligenceWikipedia

ofexceptions.Almostnothingissimplytrueorfalseinthewaythatabstractlogicrequires.AIresearch
hasexploredanumberofsolutionstothisproblem.[51]

Thebreadthofcommonsenseknowledge
Thenumberofatomicfactsthattheaveragepersonknowsisverylarge.Researchprojectsthatattemptto
buildacompleteknowledgebaseofcommonsenseknowledge(e.g.,Cyc)requireenormousamountsof
laboriousontologicalengineeringtheymustbebuilt,byhand,onecomplicatedconceptatatime.[52]A
majorgoalistohavethecomputerunderstandenoughconceptstobeabletolearnbyreadingfrom
sourcesliketheInternet,andthusbeabletoaddtoitsownontology.

Thesubsymbolicformofsomecommonsenseknowledge
Muchofwhatpeopleknowisnotrepresentedas"facts"or"statements"thattheycouldexpressverbally.
Forexample,achessmasterwillavoidaparticularchesspositionbecauseit"feelstooexposed"[53]oran
artcriticcantakeonelookatastatueandrealizethatitisafake.[54]Thesearenonconsciousandsub
symbolicintuitionsortendenciesinthehumanbrain.[55]Knowledgelikethisinforms,supportsand
providesacontextforsymbolic,consciousknowledge.Aswiththerelatedproblemofsubsymbolic
reasoning,itishopedthatsituatedAI,computationalintelligence,orstatisticalAIwillprovidewaysto
representthiskindofknowledge.[55]

Planning

Intelligentagentsmustbeabletosetgoalsandachievethem.[56]They
needawaytovisualizethefuturearepresentationofthestateofthe
worldandbeabletomakepredictionsabouthowtheiractionswill
changeitandbeabletomakechoicesthatmaximizetheutility(or
"value")ofavailablechoices.[57]

Inclassicalplanningproblems,theagentcanassumethatitistheonly
systemactingintheworld,allowingtheagenttobecertainofthe
consequencesofitsactions.[58]However,iftheagentisnottheonly
actor,thenitrequiresthattheagentcanreasonunderuncertainty.This
callsforanagentthatcannotonlyassessitsenvironmentandmake
predictions,butalsoevaluateitspredictionsandadaptbasedonits Ahierarchicalcontrolsystemisaform
assessment.[59] ofcontrolsysteminwhichasetof
devicesandgoverningsoftwareis
Multiagentplanningusesthecooperationandcompetitionofmany arrangedinahierarchy.
agentstoachieveagivengoal.Emergentbehaviorsuchasthisisused
byevolutionaryalgorithmsandswarmintelligence.[60]

Learning

Machinelearning,afundamentalconceptofAIresearchsincethefield'sinception,[61]isthestudyofcomputer
algorithmsthatimproveautomaticallythroughexperience.[62][63]

Unsupervisedlearningistheabilitytofindpatternsinastreamofinput.Supervisedlearningincludesboth
classificationandnumericalregression.Classificationisusedtodeterminewhatcategorysomethingbelongsin,
afterseeinganumberofexamplesofthingsfromseveralcategories.Regressionistheattempttoproducea
functionthatdescribestherelationshipbetweeninputsandoutputsandpredictshowtheoutputsshouldchange
astheinputschange.Inreinforcementlearning[64]theagentisrewardedforgoodresponsesandpunishedfor
badones.Theagentusesthissequenceofrewardsandpunishmentstoformastrategyforoperatinginits
problemspace.Thesethreetypesoflearningcanbeanalyzedintermsofdecisiontheory,usingconceptslike
utility.Themathematicalanalysisofmachinelearningalgorithmsandtheirperformanceisabranchof
theoreticalcomputerscienceknownascomputationallearningtheory.[65]

https://en.wikipedia.org/wiki/Artificial_intelligence 5/42
8/18/2017 ArtificialintelligenceWikipedia

Withindevelopmentalrobotics,developmentallearningapproachesareelaboratedupontoallowrobotsto
accumulaterepertoiresofnovelskillsthroughautonomousselfexploration,socialinteractionwithhuman
teachers,andtheuseofguidancemechanisms(activelearning,maturation,motorsynergies,etc.).[66][67][68][69]

Naturallanguageprocessing

Naturallanguageprocessing[70]givesmachinestheabilitytoreadand
understandhumanlanguage.Asufficientlypowerfulnaturallanguage
processingsystemwouldenablenaturallanguageuserinterfacesand
theacquisitionofknowledgedirectlyfromhumanwrittensources,such
asnewswiretexts.Somestraightforwardapplicationsofnatural
languageprocessingincludeinformationretrieval,textmining,question
answering[71]andmachinetranslation.[72]

Acommonmethodofprocessingandextractingmeaningfromnatural
languageisthroughsemanticindexing.Althoughtheseindexesrequire
alargevolumeofuserinput,itisexpectedthatincreasesinprocessor
speedsanddecreasesindatastoragecostswillresultingreater
efficiency.
Aparsetreerepresentsthesyntactic
structureofasentenceaccordingto
Perception someformalgrammar.

Machineperception[73]istheabilitytouseinputfromsensors(suchas
cameras,microphones,tactilesensors,sonarandothers)todeduceaspectsoftheworld.Computervision[74]is
theabilitytoanalyzevisualinput.Afewselectedsubproblemsarespeechrecognition,[75]facialrecognition
andobjectrecognition.[76]

Motionandmanipulation

Thefieldofrobotics[77]iscloselyrelatedtoAI.Intelligenceisrequiredforrobotstohandletaskssuchasobject
manipulation[78]andnavigation,withsubproblemssuchaslocalization,mapping,andmotionplanning.These
systemsrequirethatanagentisableto:Bespatiallycognizantofitssurroundings,learnfromandbuildamap
ofitsenvironment,figureouthowtogetfromonepointinspacetoanother,andexecutethatmovement(which
ofteninvolvescompliantmotion,aprocesswheremovementrequiresmaintainingphysicalcontactwithan
object).[79][80]

Socialintelligence

Affectivecomputingisthestudyanddevelopmentofsystemsthatcan
recognize,interpret,process,andsimulatehumanaffects.[82][83]Itisan
interdisciplinaryfieldspanningcomputersciences,psychology,and
cognitivescience.[84]Whiletheoriginsofthefieldmaybetracedasfar
backastheearlyphilosophicalinquiriesintoemotion,[85]themore
modernbranchofcomputerscienceoriginatedwithRosalindPicard's
1995paper[86]on"affectivecomputing".[87][88]Amotivationforthe
researchistheabilitytosimulateempathy,wherethemachinewouldbe
abletointerprethumanemotionsandadaptsitsbehaviortogivean
appropriateresponsetothoseemotions.
Kismet,arobotwithrudimentarysocial
skills[81]
Emotionandsocialskills[89]areimportanttoanintelligentagentfor
tworeasons.First,beingabletopredicttheactionsofothersby
understandingtheirmotivesandemotionalstatesallowanagentto
makebetterdecisions.Conceptssuchasgametheory,decisiontheory,necessitatethatanagentbeableto

https://en.wikipedia.org/wiki/Artificial_intelligence 6/42
8/18/2017 ArtificialintelligenceWikipedia

detectandmodelhumanemotions.Second,inanefforttofacilitatehumancomputerinteraction,anintelligent
machinemaywanttodisplayemotions(evenifitdoesnotexperiencethoseemotionsitself)toappearmore
sensitivetotheemotionaldynamicsofhumaninteraction.

Creativity

AsubfieldofAIaddressescreativityboththeoretically(thephilosophicalpsychologicalperspective)and
practically(thespecificimplementationofsystemsthatgeneratenovelandusefuloutputs).

Generalintelligence

Manyresearchersthinkthattheirworkwilleventuallybeincorporatedintoamachinewithartificialgeneral
intelligence,combiningalltheskillsmentionedaboveandevenexceedinghumanabilityinmostorallthese
areas.[9][90]Afewbelievethatanthropomorphicfeatureslikeartificialconsciousnessoranartificialbrainmay
berequiredforsuchaproject.[91][92]

Manyoftheproblemsabovealsorequirethatgeneralintelligencebesolved.Forexample,evenspecific
straightforwardtasks,likemachinetranslation,requirethatamachinereadandwriteinbothlanguages(NLP),
followtheauthor'sargument(reason),knowwhatisbeingtalkedabout(knowledge),andfaithfullyreproduce
theauthor'soriginalintent(socialintelligence).Aproblemlikemachinetranslationisconsidered"AI
complete",butalloftheseproblemsneedtobesolvedsimultaneouslyinordertoreachhumanlevelmachine
performance.

Approaches
ThereisnoestablishedunifyingtheoryorparadigmthatguidesAIresearch.Researchersdisagreeaboutmany
issues.[93]Afewofthemostlongstandingquestionsthathaveremainedunansweredarethese:shouldartificial
intelligencesimulatenaturalintelligencebystudyingpsychologyorneurology?Orishumanbiologyas
irrelevanttoAIresearchasbirdbiologyistoaeronauticalengineering?[94]Canintelligentbehaviorbe
describedusingsimple,elegantprinciples(suchaslogicoroptimization)?Ordoesitnecessarilyrequire
solvingalargenumberofcompletelyunrelatedproblems?[95]Canintelligencebereproducedusinghighlevel
symbols,similartowordsandideas?Ordoesitrequire"subsymbolic"processing?[96]JohnHaugeland,who
coinedthetermGOFAI(GoodOldFashionedArtificialIntelligence),alsoproposedthatAIshouldmore
properlybereferredtoassyntheticintelligence,[97]atermwhichhassincebeenadoptedbysomenonGOFAI
researchers.[98][99]

StuartShapirodividesAIresearchintothreeapproaches,whichhecallscomputationalpsychology,
computationalphilosophy,andcomputerscience.Computationalpsychologyisusedtomakecomputer
programsthatmimichumanbehavior.[100]Computationalphilosophy,isusedtodevelopanadaptive,free
flowingcomputermind.[100]Implementingcomputerscienceservesthegoalofcreatingcomputersthatcan
performtasksthatonlypeoplecouldpreviouslyaccomplish.[100]Together,thehumanesquebehavior,mind,
andactionsmakeupartificialintelligence.

Cyberneticsandbrainsimulation

Inthe1940sand1950s,anumberofresearchersexploredtheconnectionbetweenneurology,information
theory,andcybernetics.Someofthembuiltmachinesthatusedelectronicnetworkstoexhibitrudimentary
intelligence,suchasW.GreyWalter'sturtlesandtheJohnsHopkinsBeast.Manyoftheseresearchersgathered
formeetingsoftheTeleologicalSocietyatPrincetonUniversityandtheRatioClubinEngland.[18]By1960,
thisapproachwaslargelyabandoned,althoughelementsofitwouldberevivedinthe1980s.

Symbolic

https://en.wikipedia.org/wiki/Artificial_intelligence 7/42
8/18/2017 ArtificialintelligenceWikipedia

Whenaccesstodigitalcomputersbecamepossibleinthemiddle1950s,AIresearchbegantoexplorethe
possibilitythathumanintelligencecouldbereducedtosymbolmanipulation.Theresearchwascenteredin
threeinstitutions:CarnegieMellonUniversity,StanfordandMIT,andeachonedevelopeditsownstyleof
research.JohnHaugelandnamedtheseapproachestoAI"goodoldfashionedAI"or"GOFAI".[101]Duringthe
1960s,symbolicapproacheshadachievedgreatsuccessatsimulatinghighlevelthinkinginsmall
demonstrationprograms.Approachesbasedoncyberneticsorneuralnetworkswereabandonedorpushedinto
thebackground.[102]Researchersinthe1960sandthe1970swereconvincedthatsymbolicapproacheswould
eventuallysucceedincreatingamachinewithartificialgeneralintelligenceandconsideredthisthegoaloftheir
field.

Cognitivesimulation

EconomistHerbertSimonandAllenNewellstudiedhumanproblemsolvingskillsandattemptedtoformalize
them,andtheirworklaidthefoundationsofthefieldofartificialintelligence,aswellascognitivescience,
operationsresearchandmanagementscience.Theirresearchteamusedtheresultsofpsychological
experimentstodevelopprogramsthatsimulatedthetechniquesthatpeopleusedtosolveproblems.This
tradition,centeredatCarnegieMellonUniversitywouldeventuallyculminateinthedevelopmentoftheSoar
architectureinthemiddle1980s.[103][104]

Logicbased

UnlikeNewellandSimon,JohnMcCarthyfeltthatmachinesdidnotneedtosimulatehumanthought,but
shouldinsteadtrytofindtheessenceofabstractreasoningandproblemsolving,regardlessofwhetherpeople
usedthesamealgorithms.[94]HislaboratoryatStanford(SAIL)focusedonusingformallogictosolveawide
varietyofproblems,includingknowledgerepresentation,planningandlearning.[105]Logicwasalsothefocus
oftheworkattheUniversityofEdinburghandelsewhereinEuropewhichledtothedevelopmentofthe
programminglanguagePrologandthescienceoflogicprogramming.[106]

Antilogicorscruffy

ResearchersatMIT(suchasMarvinMinskyandSeymourPapert)[107]foundthatsolvingdifficultproblemsin
visionandnaturallanguageprocessingrequiredadhocsolutionstheyarguedthattherewasnosimpleand
generalprinciple(likelogic)thatwouldcapturealltheaspectsofintelligentbehavior.RogerSchankdescribed
their"antilogic"approachesas"scruffy"(asopposedtothe"neat"paradigmsatCMUandStanford).[95]
Commonsenseknowledgebases(suchasDougLenat'sCyc)areanexampleof"scruffy"AI,sincetheymustbe
builtbyhand,onecomplicatedconceptatatime.[108]

Knowledgebased

Whencomputerswithlargememoriesbecameavailablearound1970,researchersfromallthreetraditions
begantobuildknowledgeintoAIapplications.[109]This"knowledgerevolution"ledtothedevelopmentand
deploymentofexpertsystems(introducedbyEdwardFeigenbaum),thefirsttrulysuccessfulformofAI
software.[28]Theknowledgerevolutionwasalsodrivenbytherealizationthatenormousamountsof
knowledgewouldberequiredbymanysimpleAIapplications.

Subsymbolic

Bythe1980sprogressinsymbolicAIseemedtostallandmanybelievedthatsymbolicsystemswouldneverbe
abletoimitatealltheprocessesofhumancognition,especiallyperception,robotics,learningandpattern
recognition.Anumberofresearchersbegantolookinto"subsymbolic"approachestospecificAI
problems.[96]Subsymbolicmethodsmanagetoapproachintelligencewithoutspecificrepresentationsof
knowledge.

Embodiedintelligence
https://en.wikipedia.org/wiki/Artificial_intelligence 8/42
8/18/2017 ArtificialintelligenceWikipedia

Thisincludesembodied,situated,behaviorbased,andnouvelleAI.Researchersfromtherelatedfieldof
robotics,suchasRodneyBrooks,rejectedsymbolicAIandfocusedonthebasicengineeringproblemsthat
wouldallowrobotstomoveandsurvive.[110]Theirworkrevivedthenonsymbolicviewpointoftheearly
cyberneticsresearchersofthe1950sandreintroducedtheuseofcontroltheoryinAI.Thiscoincidedwiththe
developmentoftheembodiedmindthesisintherelatedfieldofcognitivescience:theideathataspectsofthe
body(suchasmovement,perceptionandvisualization)arerequiredforhigherintelligence.

Computationalintelligenceandsoftcomputing

Interestinneuralnetworksand"connectionism"wasrevivedbyDavidRumelhartandothersinthemiddleof
1980s.[111]Neuralnetworksareanexampleofsoftcomputingtheyaresolutionstoproblemswhichcannot
besolvedwithcompletelogicalcertainty,andwhereanapproximatesolutionisoftensufficient.Othersoft
computingapproachestoAIincludefuzzysystems,evolutionarycomputationandmanystatisticaltools.The
applicationofsoftcomputingtoAIisstudiedcollectivelybytheemergingdisciplineofcomputational
intelligence.[112]

Machinelearningandstatistics

Inthe1990s,AIresearchersdevelopedsophisticatedmathematicaltoolstosolvespecificsubproblems.These
toolsaretrulyscientific,inthesensethattheirresultsarebothmeasurableandverifiable,andtheyhavebeen
responsibleformanyofAI'srecentsuccesses.Thesharedmathematicallanguagehasalsopermittedahigh
levelofcollaborationwithmoreestablishedfields(likemathematics,economicsoroperationsresearch).Stuart
RussellandPeterNorvigdescribethismovementasnothinglessthana"revolution"and"thevictoryofthe
neats".[31]Criticsarguethatthesetechniques(withfewexceptions[113])aretoofocusedonparticularproblems
andhavefailedtoaddressthelongtermgoalofgeneralintelligence.[114]Thereisanongoingdebateaboutthe
relevanceandvalidityofstatisticalapproachesinAI,exemplifiedinpartbyexchangesbetweenPeterNorvig
andNoamChomsky.[115][116]

Integratingtheapproaches
Intelligentagentparadigm
Anintelligentagentisasystemthatperceivesitsenvironmentandtakesactionswhichmaximizeits
chancesofsuccess.Thesimplestintelligentagentsareprogramsthatsolvespecificproblems.More
complicatedagentsincludehumanbeingsandorganizationsofhumanbeings(suchasfirms).The
paradigmgivesresearcherslicensetostudyisolatedproblemsandfindsolutionsthatarebothverifiable
anduseful,withoutagreeingononesingleapproach.Anagentthatsolvesaspecificproblemcanuseany
approachthatworkssomeagentsaresymbolicandlogical,somearesubsymbolicneuralnetworksand
othersmayusenewapproaches.Theparadigmalsogivesresearchersacommonlanguageto
communicatewithotherfieldssuchasdecisiontheoryandeconomicsthatalsouseconceptsof
abstractagents.Theintelligentagentparadigmbecamewidelyacceptedduringthe1990s.[1]

Agentarchitecturesandcognitivearchitectures
Researchershavedesignedsystemstobuildintelligentsystemsoutofinteractingintelligentagentsina
multiagentsystem.[117]Asystemwithbothsymbolicandsubsymboliccomponentsisahybrid
intelligentsystem,andthestudyofsuchsystemsisartificialintelligencesystemsintegration.A
hierarchicalcontrolsystemprovidesabridgebetweensubsymbolicAIatitslowest,reactivelevelsand
traditionalsymbolicAIatitshighestlevels,whererelaxedtimeconstraintspermitplanningandworld
modelling.[118]RodneyBrooks'subsumptionarchitecturewasanearlyproposalforsuchahierarchical
system.[119]

Tools
Inthecourseof60+yearsofresearch,AIhasdevelopedalargenumberoftoolstosolvethemostdifficult
problemsincomputerscience.Afewofthemostgeneralofthesemethodsarediscussedbelow.

https://en.wikipedia.org/wiki/Artificial_intelligence 9/42
8/18/2017 ArtificialintelligenceWikipedia

Searchandoptimization

ManyproblemsinAIcanbesolvedintheorybyintelligentlysearchingthroughmanypossiblesolutions:[120]
Reasoningcanbereducedtoperformingasearch.Forexample,logicalproofcanbeviewedassearchingfora
paththatleadsfrompremisestoconclusions,whereeachstepistheapplicationofaninferencerule.[121]
Planningalgorithmssearchthroughtreesofgoalsandsubgoals,attemptingtofindapathtoatargetgoal,a
processcalledmeansendsanalysis.[122]Roboticsalgorithmsformovinglimbsandgraspingobjectsuselocal
searchesinconfigurationspace.[78]Manylearningalgorithmsusesearchalgorithmsbasedonoptimization.

Simpleexhaustivesearches[123]arerarelysufficientformostrealworldproblems:thesearchspace(the
numberofplacestosearch)quicklygrowstoastronomicalnumbers.Theresultisasearchthatistooslowor
nevercompletes.Thesolution,formanyproblems,istouse"heuristics"or"rulesofthumb"thateliminate
choicesthatareunlikelytoleadtothegoal(called"pruningthesearchtree").Heuristicssupplytheprogram
witha"bestguess"forthepathonwhichthesolutionlies.[124]Heuristicslimitthesearchforsolutionsintoa
smallersamplesize.[79]

Averydifferentkindofsearchcametoprominenceinthe1990s,basedonthemathematicaltheoryof
optimization.Formanyproblems,itispossibletobeginthesearchwithsomeformofaguessandthenrefine
theguessincrementallyuntilnomorerefinementscanbemade.Thesealgorithmscanbevisualizedasblind
hillclimbing:webeginthesearchatarandompointonthelandscape,andthen,byjumpsorsteps,wekeep
movingourguessuphill,untilwereachthetop.Otheroptimizationalgorithmsaresimulatedannealing,beam
searchandrandomoptimization.[125]

Evolutionarycomputationusesaformofoptimizationsearch.Forexample,theymaybeginwithapopulation
oforganisms(theguesses)andthenallowthemtomutateandrecombine,selectingonlythefittesttosurvive
eachgeneration(refiningtheguesses).Formsofevolutionarycomputationincludeswarmintelligence
algorithms(suchasantcolonyorparticleswarmoptimization)[126]andevolutionaryalgorithms(suchas
geneticalgorithms,geneexpressionprogramming,andgeneticprogramming).[127]

Logic

Logic[128]isusedforknowledgerepresentationandproblemsolving,butitcanbeappliedtootherproblemsas
well.Forexample,thesatplanalgorithmuseslogicforplanning[129]andinductivelogicprogrammingisa
methodforlearning.[130]

SeveraldifferentformsoflogicareusedinAIresearch.Propositionalorsententiallogic[131]isthelogicof
statementswhichcanbetrueorfalse.Firstorderlogic[132]alsoallowstheuseofquantifiersandpredicates,
andcanexpressfactsaboutobjects,theirproperties,andtheirrelationswitheachother.Fuzzylogic,[133]isa
versionoffirstorderlogicwhichallowsthetruthofastatementtoberepresentedasavaluebetween0and1,
ratherthansimplyTrue(1)orFalse(0).Fuzzysystemscanbeusedforuncertainreasoningandhavebeen
widelyusedinmodernindustrialandconsumerproductcontrolsystems.Subjectivelogic[134]models
uncertaintyinadifferentandmoreexplicitmannerthanfuzzylogic:agivenbinomialopinionsatisfiesbelief+
disbelief+uncertainty=1withinaBetadistribution.Bythismethod,ignorancecanbedistinguishedfrom
probabilisticstatementsthatanagentmakeswithhighconfidence.

Defaultlogics,nonmonotoniclogicsandcircumscription[51]areformsoflogicdesignedtohelpwithdefault
reasoningandthequalificationproblem.Severalextensionsoflogichavebeendesignedtohandlespecific
domainsofknowledge,suchas:descriptionlogics[135]situationcalculus,eventcalculusandfluentcalculus
(forrepresentingeventsandtime)[136]causalcalculus[137]beliefcalculus[138]andmodallogics.[139]

Probabilisticmethodsforuncertainreasoning

ManyproblemsinAI(inreasoning,planning,learning,perceptionandrobotics)requiretheagenttooperate
withincompleteoruncertaininformation.AIresearchershavedevisedanumberofpowerfultoolstosolve
theseproblemsusingmethodsfromprobabilitytheoryandeconomics.[140]
Bayesiannetworks [141] areaverygeneraltoolthatcanbeusedforalargenumberofproblems:reasoning
https://en.wikipedia.org/wiki/Artificial_intelligence 10/42
8/18/2017 ArtificialintelligenceWikipedia

Bayesiannetworks[141]areaverygeneraltoolthatcanbeusedforalargenumberofproblems:reasoning
(usingtheBayesianinferencealgorithm),[142]learning(usingtheexpectationmaximizationalgorithm),[143]
planning(usingdecisionnetworks)[144]andperception(usingdynamicBayesiannetworks).[145]Probabilistic
algorithmscanalsobeusedforfiltering,prediction,smoothingandfindingexplanationsforstreamsofdata,
helpingperceptionsystemstoanalyzeprocessesthatoccurovertime(e.g.,hiddenMarkovmodelsorKalman
filters).[145]

Akeyconceptfromthescienceofeconomicsis"utility":ameasureofhowvaluablesomethingistoan
intelligentagent.Precisemathematicaltoolshavebeendevelopedthatanalyzehowanagentcanmakechoices
andplan,usingdecisiontheory,decisionanalysis,[146]andinformationvaluetheory.[57]Thesetoolsinclude
modelssuchasMarkovdecisionprocesses,[147]dynamicdecisionnetworks,[145]gametheoryandmechanism
design.[148]

Classifiersandstatisticallearningmethods

ThesimplestAIapplicationscanbedividedintotwotypes:classifiers("ifshinythendiamond")andcontrollers
("ifshinythenpickup").Controllersdo,however,alsoclassifyconditionsbeforeinferringactions,and
thereforeclassificationformsacentralpartofmanyAIsystems.Classifiersarefunctionsthatusepattern
matchingtodetermineaclosestmatch.Theycanbetunedaccordingtoexamples,makingthemveryattractive
foruseinAI.Theseexamplesareknownasobservationsorpatterns.Insupervisedlearning,eachpattern
belongstoacertainpredefinedclass.Aclasscanbeseenasadecisionthathastobemade.Alltheobservations
combinedwiththeirclasslabelsareknownasadataset.Whenanewobservationisreceived,thatobservation
isclassifiedbasedonpreviousexperience.[149]

Aclassifiercanbetrainedinvariouswaystherearemanystatisticalandmachinelearningapproaches.The
mostwidelyusedclassifiersaretheneuralnetwork,[150]kernelmethodssuchasthesupportvector
machine,[151]knearestneighboralgorithm,[152]Gaussianmixturemodel,[153]naiveBayesclassifier,[154]and
decisiontree.[155]Theperformanceoftheseclassifiershavebeencomparedoverawiderangeoftasks.
Classifierperformancedependsgreatlyonthecharacteristicsofthedatatobeclassified.Thereisnosingle
classifierthatworksbestonallgivenproblemsthisisalsoreferredtoasthe"nofreelunch"theorem.
Determiningasuitableclassifierforagivenproblemisstillmoreanartthanscience.[156]

Neuralnetworks

Thestudyofnonlearningartificialneuralnetworks[150]beganinthe
decadebeforethefieldofAIresearchwasfounded,intheworkof
WalterPittsandWarrenMcCullouch.FrankRosenblattinventedthe
perceptron,alearningnetworkwithasinglelayer,similartotheold
conceptoflinearregression.EarlypioneersalsoincludeAlexey
GrigorevichIvakhnenko,TeuvoKohonen,StephenGrossberg,
KunihikoFukushima,ChristophvonderMalsburg,DavidWillshaw,
ShunIchiAmari,BernardWidrow,JohnHopfield,EduardoR.
Caianiello,andothers.

Themaincategoriesofnetworksareacyclicorfeedforwardneural
networks(wherethesignalpassesinonlyonedirection)andrecurrent Aneuralnetworkisaninterconnected
neuralnetworks(whichallowfeedbackandshorttermmemoriesof groupofnodes,akintothevast
previousinputevents).Amongthemostpopularfeedforwardnetworks networkofneuronsinthehumanbrain.
areperceptrons,multilayerperceptronsandradialbasisnetworks.[157]
Neuralnetworkscanbeappliedtotheproblemofintelligentcontrol
(forrobotics)orlearning,usingsuchtechniquesasHebbianlearning,GMDHorcompetitivelearning.[158]

Today,neuralnetworksareoftentrainedbythebackpropagationalgorithm,whichhadbeenaroundsince1970
asthereversemodeofautomaticdifferentiationpublishedbySeppoLinnainmaa,[159][160]andwasintroduced
toneuralnetworksbyPaulWerbos.[161][162][163]
https://en.wikipedia.org/wiki/Artificial_intelligence 11/42
8/18/2017 ArtificialintelligenceWikipedia

Hierarchicaltemporalmemoryisanapproachthatmodelssomeofthestructuralandalgorithmicpropertiesof
theneocortex.[164]

Deepfeedforwardneuralnetworks

Deeplearninginartificialneuralnetworkswithmanylayershastransformedmanyimportantsubfieldsof
artificialintelligence,includingcomputervision,speechrecognition,naturallanguageprocessingand
others.[165][166][167]

Accordingtoasurvey,[168]theexpression"DeepLearning"wasintroducedtotheMachineLearning
communitybyRinaDechterin1986[169]andgainedtractionafterIgorAizenbergandcolleaguesintroducedit
toArtificialNeuralNetworksin2000.[170]ThefirstfunctionalDeepLearningnetworkswerepublishedby
AlexeyGrigorevichIvakhnenkoandV.G.Lapain1965.[171]Thesenetworksaretrainedonelayeratatime.
Ivakhnenko's1971paper[172]describesthelearningofadeepfeedforwardmultilayerperceptronwitheight
layers,alreadymuchdeeperthanmanylaternetworks.In2006,apublicationbyGeoffreyHintonandRuslan
Salakhutdinovintroducedanotherwayofpretrainingmanylayeredfeedforwardneuralnetworks(FNNs)one
layeratatime,treatingeachlayerinturnasanunsupervisedrestrictedBoltzmannmachine,thenusing
supervisedbackpropagationforfinetuning.[173]Similartoshallowartificialneuralnetworks,deepneural
networkscanmodelcomplexnonlinearrelationships.Overthelastfewyears,advancesinbothmachine
learningalgorithmsandcomputerhardwarehaveledtomoreefficientmethodsfortrainingdeepneural
networksthatcontainmanylayersofnonlinearhiddenunitsandaverylargeoutputlayer.[174]

Deeplearningoftenusesconvolutionalneuralnetworks(CNNs),whoseoriginscanbetracedbacktothe
NeocognitronintroducedbyKunihikoFukushimain1980.[175]In1989,YannLeCunandcolleaguesapplied
backpropagationtosuchanarchitecture.Intheearly2000s,inanindustrialapplicationCNNsalready
processedanestimated10%to20%ofallthecheckswrittenintheUS.[176]Since2011,fastimplementations
ofCNNsonGPUshavewonmanyvisualpatternrecognitioncompetitions.[167]

DeepfeedforwardneuralnetworkswereusedinconjunctionwithreinforcementlearningbyAlphaGo,Google
Deepmind'sprogramthatwasthefirsttobeataprofessionalhumanGoplayer.[177]

Deeprecurrentneuralnetworks

Earlyon,deeplearningwasalsoappliedtosequencelearningwithrecurrentneuralnetworks(RNNs)[178]
whicharegeneralcomputersandcanrunarbitraryprogramstoprocessarbitrarysequencesofinputs.Thedepth
ofanRNNisunlimitedanddependsonthelengthofitsinputsequence.[167]RNNscanbetrainedbygradient
descent[179][180][181]butsufferfromthevanishinggradientproblem.[165][182]In1992,itwasshownthat
unsupervisedpretrainingofastackofrecurrentneuralnetworkscanspeedupsubsequentsupervisedlearning
ofdeepsequentialproblems.[183]

NumerousresearchersnowusevariantsofadeeplearningrecurrentNNcalledthelongshorttermmemory
(LSTM)networkpublishedbyHochreiter&Schmidhuberin1997.[184]LSTMisoftentrainedby
ConnectionistTemporalClassification(CTC).[185]AtGoogle,MicrosoftandBaiduthisapproachhas
revolutionisedspeechrecognition.[186][187][188]Forexample,in2015,Google'sspeechrecognition
experiencedadramaticperformancejumpof49%throughCTCtrainedLSTM,whichisnowavailablethrough
GoogleVoicetobillionsofsmartphoneusers.[189]GooglealsousedLSTMtoimprovemachine
translation,[190]LanguageModeling[191]andMultilingualLanguageProcessing.[192]LSTMcombinedwith
CNNsalsoimprovedautomaticimagecaptioning[193]andaplethoraofotherapplications.

Controltheory

Controltheory,thegrandchildofcybernetics,hasmanyimportantapplications,especiallyinrobotics.[194]

Languages

AIresearchershavedevelopedseveralspecializedlanguagesforAIresearch,includingLisp
https://en.wikipedia.org/wiki/Artificial_intelligence
[195] and 12/42
8/18/2017 ArtificialintelligenceWikipedia

AIresearchershavedevelopedseveralspecializedlanguagesforAIresearch,includingLisp[195]and
Prolog.[196]

Evaluatingprogress

In1950,AlanTuringproposedageneralproceduretotesttheintelligenceofanagentnowknownastheTuring
test.Thisprocedureallowsalmostallthemajorproblemsofartificialintelligencetobetested.However,itisa
verydifficultchallengeandatpresentallagentsfail.[197]

Artificialintelligencecanalsobeevaluatedonspecificproblemssuchassmallproblemsinchemistry,hand
writingrecognitionandgameplaying.SuchtestshavebeentermedsubjectmatterexpertTuringtests.Smaller
problemsprovidemoreachievablegoalsandthereareaneverincreasingnumberofpositiveresults.[198]

Forexample,performanceatdraughts(i.e.checkers)isoptimal,[199]performanceatchessishighhumanand
nearingsuperhuman(seecomputerchess:computersversushuman)andperformanceatmanyeverydaytasks
(suchasrecognizingafaceorcrossingaroomwithoutbumpingintosomething)issubhuman.

Aquitedifferentapproachmeasuresmachineintelligencethroughtestswhicharedevelopedfrom
mathematicaldefinitionsofintelligence.Examplesofthesekindsoftestsstartinthelateninetiesdevising
intelligencetestsusingnotionsfromKolmogorovcomplexityanddatacompression.[200]Twomajor
advantagesofmathematicaldefinitionsaretheirapplicabilitytononhumanintelligencesandtheirabsenceofa
requirementforhumantesters.

AderivativeoftheTuringtestistheCompletelyAutomatedPublicTuringtesttotellComputersandHumans
Apart(CAPTCHA).Asthenameimplies,thishelpstodeterminethatauserisanactualpersonandnota
computerposingasahuman.IncontrasttothestandardTuringtest,CAPTCHAadministeredbyamachineand
targetedtoahumanasopposedtobeingadministeredbyahumanandtargetedtoamachine.Acomputerasksa
usertocompleteasimpletestthengeneratesagradeforthattest.Computersareunabletosolvetheproblem,
socorrectsolutionsaredeemedtobetheresultofapersontakingthetest.AcommontypeofCAPTCHAisthe
testthatrequiresthetypingofdistortedletters,numbersorsymbolsthatappearinanimageundecipherableby
acomputer.[201]

Applications
AIisrelevanttoanyintellectualtask.[202]Modernartificialintelligence
techniquesarepervasiveandaretoonumeroustolisthere.Frequently,
whenatechniquereachesmainstreamuse,itisnolongerconsidered
artificialintelligencethisphenomenonisdescribedastheAI
effect.[203]

HighprofileexamplesofAIincludeautonomousvehicles(suchas
dronesandselfdrivingcars),medicaldiagnosis,creatingart(suchas
poetry),provingmathematicaltheorems,playinggames(suchasChess
orGo),searchengines(suchasGooglesearch),onlineassistants(such
asSiri),imagerecognitioninphotographs,spamfiltering,predictionof
judicialdecisions[204]andtargetingonline
advertisements.[202][205][206]

WithsocialmediasitesovertakingTVasasourcefornewsforyoung
peopleandnewsorganisationsincreasinglyreliantonsocialmedia
Anautomatedonlineassistant
platformsforgeneratingdistribution,[207]majorpublishersnowuse
providingcustomerserviceonaweb
artificialintelligence(AI)technologytopoststoriesmoreeffectively
pageoneofmanyveryprimitive
andgeneratehighervolumesoftraffic.[208]
applicationsofartificialintelligence.

Competitionsandprizes
https://en.wikipedia.org/wiki/Artificial_intelligence 13/42
8/18/2017 ArtificialintelligenceWikipedia

Thereareanumberofcompetitionsandprizestopromoteresearchinartificialintelligence.Themainareas
promotedare:generalmachineintelligence,conversationalbehavior,datamining,roboticcars,robotsoccer
andgames.

Healthcare

Artificialintelligenceisbreakingintothehealthcareindustrybyassistingdoctors.AccordingtoBloomberg
Technology,MicrosofthasdevelopedAItohelpdoctorsfindtherighttreatmentsforcancer.[209]Thereisa
greatamountofresearchanddrugsdevelopedrelatingtocancer.Indetail,therearemorethan800medicines
andvaccinestotreatcancer.Thisnegativelyaffectsthedoctors,becausetherearewaytoomanyoptionsto
choosefrom,makingitmoredifficulttochoosetherightdrugsforthepatients.Microsoftisworkingona
projecttodevelopamachinecalled"Hanover".Itsgoalistomemorizeallthepapersnecessarytocancerand
helppredictwhichcombinationsofdrugswillbemosteffectiveforeachpatient.Oneprojectthatisbeing
workedonatthemomentisfightingmyeloidleukemia,afatalcancerwherethetreatmenthasnotimprovedin
decades.Anotherstudywasreportedtohavefoundthatartificialintelligencewasasgoodastraineddoctorsin
identifyingskincancers.[210]Anotherstudyisusingartificialintelligencetotryandmonitormultiplehighrisk
patients,andthisisdonebyaskingeachpatientnumerousquestionsbasedondataacquiredfromlivedoctorto
patientinteractions.[211]

AccordingtoCNN,therewasarecentstudybysurgeonsattheChildren'sNationalMedicalCenterin
Washingtonwhichsuccessfullydemonstratedsurgerywithanautonomousrobot.Theteamsupervisedtherobot
whileitperformedsofttissuesurgery,stitchingtogetherapig'sbowelduringopensurgery,anddoingsobetter
thanahumansurgeon,theteamclaimed.[212]

Automotive

AdvancementsinAIhavecontributedtothegrowthoftheautomotiveindustrythroughthecreationand
evolutionofselfdrivingvehicles.Asof2016,thereareover30companiesutilizingAIintothecreationof
driverlesscars.AfewcompaniesinvolvedwithAIincludeTesla,Google,andApple.[213]

Manycomponentscontributetothefunctioningofselfdrivingcars.Thesevehiclesincorporatesystemssuchas
braking,lanechanging,collisionprevention,navigationandmapping.Together,thesesystems,aswellashigh
performancecomputersareintegratedintoonecomplexvehicle.[214]

Onemainfactorthatinfluencestheabilityforadriverlesscartofunctionismapping.Ingeneral,thevehicle
wouldbepreprogrammedwithamapoftheareabeingdriven.Thismapwouldincludedataonthe
approximationsofstreetlightandcurbheightsinorderforthevehicletobeawareofitssurroundings.
However,Googlehasbeenworkingonanalgorithmwiththepurposeofeliminatingtheneedforpre
programmedmapsandinstead,creatingadevicethatwouldbeabletoadjusttoavarietyofnew
surroundings.[215]Someselfdrivingcarsarenotequippedwithsteeringwheelsorbrakes,sotherehasalso
beenresearchfocusedoncreatinganalgorithmthatiscapableofmaintainingasafeenvironmentforthe
passengersinthevehiclethroughawarenessofspeedanddrivingconditions.[216]

Finance

Financialinstitutionshavelongusedartificialneuralnetworksystemstodetectchargesorclaimsoutsideofthe
norm,flaggingtheseforhumaninvestigation.

UseofAIinbankingcanbetrackedbackto1987whenSecurityPacificNationalBankinUSAsetupaFraud
PreventionTaskforcetocountertheunauthoriseduseofdebitcards.AppslikeKasisitoandMoneystreamare
usingAIinfinancialservices

Banksuseartificialintelligencesystemstoorganizeoperations,maintainbookkeeping,investinstocks,and
manageproperties.AIcanreacttochangesovernightorwhenbusinessisnottakingplace.[217]InAugust2001,
robotsbeathumansinasimulatedfinancialtradingcompetition.[218]
https://en.wikipedia.org/wiki/Artificial_intelligence 14/42
8/18/2017 ArtificialintelligenceWikipedia

AIhasalsoreducedfraudandcrimebymonitoringbehavioralpatternsofusersforanychangesor
anomalies.[219]

Videogames

Artificialintelligenceisusedtogenerateintelligentbehaviorsprimarilyinnonplayercharacters(NPCs),often
simulatinghumanlikeintelligence.[220]

Platforms
Aplatform(or"computingplatform")isdefinedas"somesortofhardwarearchitectureorsoftwareframework
(includingapplicationframeworks),thatallowssoftwaretorun".AsRodneyBrookspointedoutmanyyears
ago,[221]itisnotjusttheartificialintelligencesoftwarethatdefinestheAIfeaturesoftheplatform,butrather
theactualplatformitselfthataffectstheAIthatresults,i.e.,thereneedstobeworkinAIproblemsonreal
worldplatformsratherthaninisolation.

AwidevarietyofplatformshasalloweddifferentaspectsofAItodevelop,rangingfromexpertsystemssuch
asCyctodeeplearningframeworkstorobotplatformssuchastheRoombawithopeninterface.[222]Recent
advancesindeepartificialneuralnetworksanddistributedcomputinghaveledtoaproliferationofsoftware
libraries,includingDeeplearning4j,TensorFlow,TheanoandTorch.

CollectiveAIisaplatformarchitecturethatcombinesindividualAIintoacollectiveentity,inordertoachieve
globalresultsfromindividualbehaviors.[223][224]Withitscollectivestructure,developerscancrowdsource
informationandextendthefunctionalityofexistingAIdomainsontheplatformfortheirownuse,aswellas
continuetocreateandsharenewdomainsandcapabilitiesforthewidercommunityandgreatergood.[225]As
developerscontinuetocontribute,theoverallplatformgrowsmoreintelligentandisabletoperformmore
requests,providingascalablemodelforgreatercommunalbenefit.[224]OrganizationslikeSoundHoundInc.
andtheHarvardJohnA.PaulsonSchoolofEngineeringandAppliedScienceshaveusedthiscollaborativeAI
model.[226][224]

PartnershiponAI

Amazon,Google,Facebook,IBM,andMicrosofthaveestablishedanonprofitpartnershiptoformulatebest
practicesonartificialintelligencetechnologies,advancethepublic'sunderstanding,andtoserveasaplatform
aboutartificialintelligence.[227]Theystated:"ThispartnershiponAIwillconductresearch,organize
discussions,providethoughtleadership,consultwithrelevantthirdparties,respondtoquestionsfromthe
publicandmedia,andcreateeducationalmaterialthatadvancetheunderstandingofAItechnologiesincluding
machineperception,learning,andautomatedreasoning."[227]Applejoinedothertechcompaniesasafounding
memberofthePartnershiponAIinJanuary2017.Thecorporatememberswillmakefinancialandresearch
contributionstothegroup,whileengagingwiththescientificcommunitytobringacademicsontothe
board.[228][224]

Philosophyandethics
TherearethreephilosophicalquestionsrelatedtoAI:

1.Isartificialgeneralintelligencepossible?Canamachinesolveanyproblemthatahumanbeingcansolve
usingintelligence?Oraretherehardlimitstowhatamachinecanaccomplish?
2.Areintelligentmachinesdangerous?Howcanweensurethatmachinesbehaveethicallyandthattheyare
usedethically?
3.Canamachinehaveamind,consciousnessandmentalstatesinexactlythesamesensethathuman
beingsdo?Canamachinebesentient,andthusdeservecertainrights?Canamachineintentionallycause
harm?

https://en.wikipedia.org/wiki/Artificial_intelligence 15/42
8/18/2017 ArtificialintelligenceWikipedia

Thelimitsofartificialgeneralintelligence

Canamachinebeintelligent?Canit"think"?

Turing's"politeconvention"
Weneednotdecideifamachinecan"think"weneedonlydecideifamachinecanactasintelligentlyas
ahumanbeing.Thisapproachtothephilosophicalproblemsassociatedwithartificialintelligenceforms
thebasisoftheTuringtest.[197]

TheDartmouthproposal
"Everyaspectoflearningoranyotherfeatureofintelligencecanbesopreciselydescribedthatamachine
canbemadetosimulateit."ThisconjecturewasprintedintheproposalfortheDartmouthConferenceof
1956,andrepresentsthepositionofmostworkingAIresearchers.[229]

NewellandSimon'sphysicalsymbolsystemhypothesis
"Aphysicalsymbolsystemhasthenecessaryandsufficientmeansofgeneralintelligentaction."Newell
andSimonarguethatintelligenceconsistsofformaloperationsonsymbols.[230]HubertDreyfusargued
that,onthecontrary,humanexpertisedependsonunconsciousinstinctratherthanconscioussymbol
manipulationandonhavinga"feel"forthesituationratherthanexplicitsymbolicknowledge.(See
Dreyfus'critiqueofAI.)[231][232]

Gdelianarguments
Gdelhimself,[233]JohnLucas(in1961)andRogerPenrose(inamoredetailedargumentfrom1989
onwards)madehighlytechnicalargumentsthathumanmathematicianscanconsistentlyseethetruthof
theirown"Gdelstatements"andthereforehavecomputationalabilitiesbeyondthatofmechanical
Turingmachines.[234]However,themodernconsensusinthescientificandmathematicalcommunityis
thatthese"Gdelianarguments"fail.[235][236][237]

Theartificialbrainargument
Thebraincanbesimulatedbymachinesandbecausebrainsareintelligent,simulatedbrainsmustalsobe
intelligentthusmachinescanbeintelligent.HansMoravec,RayKurzweilandothershavearguedthatit
istechnologicallyfeasibletocopythebraindirectlyintohardwareandsoftware,andthatsucha
simulationwillbeessentiallyidenticaltotheoriginal.[92]

TheAIeffect
Machinesarealreadyintelligent,butobservershavefailedtorecognizeit.WhenDeepBluebeatGarry
Kasparovinchess,themachinewasactingintelligently.However,onlookerscommonlydiscountthe
behaviorofanartificialintelligenceprogrambyarguingthatitisnot"real"intelligenceafterallthus
"real"intelligenceiswhateverintelligentbehaviorpeoplecandothatmachinesstillcannot.Thisis
knownastheAIEffect:"AIiswhateverhasn'tbeendoneyet."

Potentialrisksandmoralreasoning

Widespreaduseofartificialintelligencecouldhaveunintendedconsequencesthataredangerousorundesirable.
ScientistsfromtheFutureofLifeInstitute,amongothers,describedsomeshorttermresearchgoalstobehow
AIinfluencestheeconomy,thelawsandethicsthatareinvolvedwithAIandhowtominimizeAIsecurity
risks.Inthelongterm,thescientistshaveproposedtocontinueoptimizingfunctionwhileminimizingpossible
securityrisksthatcomealongwithnewtechnologies.[238]

Machineswithintelligencehavethepotentialtousetheirintelligencetomakeethicaldecisions.Researchin
thisareaincludes"machineethics","artificialmoralagents",andthestudyof"malevolentvs.friendlyAI".

Existentialrisk

https://en.wikipedia.org/wiki/Artificial_intelligence 16/42
8/18/2017 ArtificialintelligenceWikipedia

Thedevelopmentoffullartificialintelligencecouldspelltheendofthehumanrace.Oncehumans
developartificialintelligence,itwilltakeoffonitsownandredesignitselfataneverincreasing
rate.Humans,whoarelimitedbyslowbiologicalevolution,couldn'tcompeteandwouldbe
superseded.

StephenHawking[239]

Acommonconcernaboutthedevelopmentofartificialintelligenceisthepotentialthreatitcouldposeto
mankind.ThisconcernhasrecentlygainedattentionaftermentionsbycelebritiesincludingStephenHawking,
BillGates,[240]andElonMusk.[241]AgroupofprominenttechtitansincludingPeterThiel,AmazonWeb
ServicesandMuskhavecommitted$1billiontoOpenAIanonprofitcompanyaimedatchampioning
responsibleAIdevelopment.[242]Theopinionofexpertswithinthefieldofartificialintelligenceismixed,with
sizablefractionsbothconcernedandunconcernedbyriskfromeventualsuperhumanlycapableAI.[243]

InhisbookSuperintelligence,NickBostromprovidesanargumentthatartificialintelligencewillposeathreat
tomankind.HearguesthatsufficientlyintelligentAI,ifitchoosesactionsbasedonachievingsomegoal,will
exhibitconvergentbehaviorsuchasacquiringresourcesorprotectingitselffrombeingshutdown.IfthisAI's
goalsdonotreflecthumanity'soneexampleisanAItoldtocomputeasmanydigitsofpiaspossibleit
mightharmhumanityinordertoacquiremoreresourcesorpreventitselffrombeingshutdown,ultimatelyto
betterachieveitsgoal.

Forthisdangertoberealized,thehypotheticalAIwouldhavetooverpoweroroutthinkallofhumanity,which
aminorityofexpertsargueisapossibilityfarenoughinthefuturetonotbeworthresearching.[244][245]Other
counterargumentsrevolvearoundhumansbeingeitherintrinsicallyorconvergentlyvaluablefromthe
perspectiveofanartificialintelligence.[246]

Concernoverriskfromartificialintelligencehasledtosomehighprofiledonationsandinvestments.In
January2015,ElonMuskdonatedtenmilliondollarstotheFutureofLifeInstitutetofundresearchon
understandingAIdecisionmaking.Thegoaloftheinstituteisto"growwisdomwithwhichwemanage"the
growingpoweroftechnology.MuskalsofundscompaniesdevelopingartificialintelligencesuchasGoogle
DeepMindandVicariousto"justkeepaneyeonwhat'sgoingonwithartificialintelligence.[247]Ithinkthereis
potentiallyadangerousoutcomethere."[248][249]

Developmentofmilitarizedartificialintelligenceisarelatedconcern.Currently,50+countriesareresearching
battlefieldrobots,includingtheUnitedStates,China,Russia,andtheUnitedKingdom.Manypeopleconcerned
aboutriskfromsuperintelligentAIalsowanttolimittheuseofartificialsoldiers.[250]

Devaluationofhumanity

JosephWeizenbaumwrotethatAIapplicationscannot,bydefinition,successfullysimulategenuinehuman
empathyandthattheuseofAItechnologyinfieldssuchascustomerserviceorpsychotherapy[251]wasdeeply
misguided.WeizenbaumwasalsobotheredthatAIresearchers(andsomephilosophers)werewillingtoview
thehumanmindasnothingmorethanacomputerprogram(apositionnowknownascomputationalism).To
WeizenbaumthesepointssuggestthatAIresearchdevalueshumanlife.[252]

Decreaseindemandforhumanlabor

MartinFord,authorofTheLightsintheTunnel:Automation,AcceleratingTechnologyandtheEconomyofthe
Future,[253]andothersarguethatspecializedartificialintelligenceapplications,roboticsandotherformsof
automationwillultimatelyresultinsignificantunemploymentasmachinesbegintomatchandexceedthe
capabilityofworkerstoperformmostroutineandrepetitivejobs.Fordpredictsthatmanyknowledgebased
occupationsandinparticularentryleveljobswillbeincreasinglysusceptibletoautomationviaexpert
systems,machinelearning[254]andotherAIenhancedapplications.AIbasedapplicationsmayalsobeusedto
amplifythecapabilitiesoflowwageoffshoreworkers,makingitmorefeasibletooutsourceknowledge
work.[255]
https://en.wikipedia.org/wiki/Artificial_intelligence 17/42
8/18/2017 ArtificialintelligenceWikipedia

Artificialmoralagents

ThisraisestheissueofhowethicallythemachineshouldbehavetowardsbothhumansandotherAIagents.
ThisissuewasaddressedbyWendellWallachinhisbooktitledMoralMachinesinwhichheintroducedthe
conceptofartificialmoralagents(AMA).[256]ForWallach,AMAshavebecomeapartoftheresearch
landscapeofartificialintelligenceasguidedbyitstwocentralquestionswhichheidentifiesas"DoesHumanity
WantComputersMakingMoralDecisions"[257]and"Can(Ro)botsReallyBeMoral".[258]ForWallachthe
questionisnotcenteredontheissueofwhethermachinescandemonstratetheequivalentofmoralbehaviorin
contrasttotheconstraintswhichsocietymayplaceonthedevelopmentofAMAs.[259]

Machineethics

Thefieldofmachineethicsisconcernedwithgivingmachinesethicalprinciples,oraprocedurefordiscovering
awaytoresolvetheethicaldilemmastheymightencounter,enablingthemtofunctioninanethically
responsiblemannerthroughtheirownethicaldecisionmaking.[260]ThefieldwasdelineatedintheAAAIFall
2005SymposiumonMachineEthics:"Pastresearchconcerningtherelationshipbetweentechnologyandethics
haslargelyfocusedonresponsibleandirresponsibleuseoftechnologybyhumanbeings,withafewpeople
beinginterestedinhowhumanbeingsoughttotreatmachines.Inallcases,onlyhumanbeingshaveengagedin
ethicalreasoning.Thetimehascomeforaddinganethicaldimensiontoatleastsomemachines.Recognitionof
theethicalramificationsofbehaviorinvolvingmachines,aswellasrecentandpotentialdevelopmentsin
machineautonomy,necessitatethis.Incontrasttocomputerhacking,softwarepropertyissues,privacyissues
andothertopicsnormallyascribedtocomputerethics,machineethicsisconcernedwiththebehaviorof
machinestowardshumanusersandothermachines.Researchinmachineethicsiskeytoalleviatingconcerns
withautonomoussystemsitcouldbearguedthatthenotionofautonomousmachineswithoutsucha
dimensionisattherootofallfearconcerningmachineintelligence.Further,investigationofmachineethics
couldenablethediscoveryofproblemswithcurrentethicaltheories,advancingourthinkingaboutEthics."[261]
Machineethicsissometimesreferredtoasmachinemorality,computationalethicsorcomputationalmorality.
Avarietyofperspectivesofthisnascentfieldcanbefoundinthecollectededition"MachineEthics"[260]that
stemsfromtheAAAIFall2005SymposiumonMachineEthics.[261]

MalevolentandfriendlyAI

PoliticalscientistCharlesT.RubinbelievesthatAIcanbeneitherdesignednorguaranteedtobe
benevolent.[262]Hearguesthat"anysufficientlyadvancedbenevolencemaybeindistinguishablefrom
malevolence."Humansshouldnotassumemachinesorrobotswouldtreatusfavorably,becausethereisnoa
priorireasontobelievethattheywouldbesympathetictooursystemofmorality,whichhasevolvedalong
withourparticularbiology(whichAIswouldnotshare).Hyperintelligentsoftwaremaynotnecessarilydecide
tosupportthecontinuedexistenceofmankind,andwouldbeextremelydifficulttostop.Thistopichasalso
recentlybeguntobediscussedinacademicpublicationsasarealsourceofriskstocivilization,humans,and
planetEarth.

PhysicistStephenHawking,MicrosoftfounderBillGates,andSpaceXfounderElonMuskhaveexpressed
concernsaboutthepossibilitythatAIcouldevolvetothepointthathumanscouldnotcontrolit,withHawking
theorizingthatthiscould"spelltheendofthehumanrace".[263]

OneproposaltodealwiththisistoensurethatthefirstgenerallyintelligentAIis'FriendlyAI',andwillthenbe
abletocontrolsubsequentlydevelopedAIs.Somequestionwhetherthiskindofcheckcouldreallyremainin
place.

LeadingAIresearcherRodneyBrookswrites,"Ithinkitisamistaketobeworryingaboutusdeveloping
malevolentAIanytimeinthenextfewhundredyears.Ithinktheworrystemsfromafundamentalerrorinnot
distinguishingthedifferencebetweentheveryrealrecentadvancesinaparticularaspectofAI,andthe
enormityandcomplexityofbuildingsentientvolitionalintelligence."[264]

https://en.wikipedia.org/wiki/Artificial_intelligence 18/42
8/18/2017 ArtificialintelligenceWikipedia

Machineconsciousness,sentienceandmind

IfanAIsystemreplicatesallkeyaspectsofhumanintelligence,willthatsystemalsobesentientwillithavea
mindwhichhasconsciousexperiences?Thisquestioniscloselyrelatedtothephilosophicalproblemastothe
natureofhumanconsciousness,generallyreferredtoasthehardproblemofconsciousness.

Consciousness

Computationalismandfunctionalism

Computationalismisthepositioninthephilosophyofmindthatthehumanmindorthehumanbrain(orboth)
isaninformationprocessingsystemandthatthinkingisaformofcomputing.[265]Computationalismargues
thattherelationshipbetweenmindandbodyissimilaroridenticaltotherelationshipbetweensoftwareand
hardwareandthusmaybeasolutiontothemindbodyproblem.Thisphilosophicalpositionwasinspiredbythe
workofAIresearchersandcognitivescientistsinthe1960sandwasoriginallyproposedbyphilosophersJerry
FodorandHilaryPutnam.

StrongAIhypothesis

ThephilosophicalpositionthatJohnSearlehasnamed"strongAI"states:"Theappropriatelyprogrammed
computerwiththerightinputsandoutputswouldtherebyhaveamindinexactlythesamesensehumanbeings
haveminds."[266]SearlecountersthisassertionwithhisChineseroomargument,whichasksustolookinside
thecomputerandtrytofindwherethe"mind"mightbe.[267]

Robotrights

MaryShelley'sFrankensteinconsidersakeyissueintheethicsofartificialintelligence:ifamachinecanbe
createdthathasintelligence,coulditalsofeel?Ifitcanfeel,doesithavethesamerightsasahuman?Theidea
alsoappearsinmodernsciencefiction,suchasthefilmA.I.:ArtificialIntelligence,inwhichhumanoid
machineshavetheabilitytofeelemotions.Thisissue,nowknownas"robotrights",iscurrentlybeing
consideredby,forexample,California'sInstitutefortheFuture,althoughmanycriticsbelievethatthe
discussionispremature.[268]Somecriticsoftranshumanismarguethatanyhypotheticalrobotrightswouldlie
onaspectrumwithanimalrightsandhumanrights.[269]Thesubjectisprofoundlydiscussedinthe2010
documentaryfilmPlug&Pray.[270]

Superintelligence

Aretherelimitstohowintelligentmachinesorhumanmachinehybridscanbe?Asuperintelligence,
hyperintelligence,orsuperhumanintelligenceisahypotheticalagentthatwouldpossessintelligencefar
surpassingthatofthebrightestandmostgiftedhumanmind.Superintelligencemayalsorefertotheformor
degreeofintelligencepossessedbysuchanagent.[90]

Technologicalsingularity

IfresearchintoStrongAIproducedsufficientlyintelligentsoftware,itmightbeabletoreprogramandimprove
itself.Theimprovedsoftwarewouldbeevenbetteratimprovingitself,leadingtorecursiveself
improvement.[271]Thenewintelligencecouldthusincreaseexponentiallyanddramaticallysurpasshumans.
SciencefictionwriterVernorVingenamedthisscenario"singularity".[272]Technologicalsingularityiswhen
acceleratingprogressintechnologieswillcausearunawayeffectwhereinartificialintelligencewillexceed
humanintellectualcapacityandcontrol,thusradicallychangingorevenendingcivilization.Becausethe
capabilitiesofsuchanintelligencemaybeimpossibletocomprehend,thetechnologicalsingularityisan
occurrencebeyondwhicheventsareunpredictableorevenunfathomable.[272][90]

https://en.wikipedia.org/wiki/Artificial_intelligence 19/42
8/18/2017 ArtificialintelligenceWikipedia

RayKurzweilhasusedMoore'slaw(whichdescribestherelentlessexponentialimprovementindigital
technology)tocalculatethatdesktopcomputerswillhavethesameprocessingpowerashumanbrainsbythe
year2029,andpredictsthatthesingularitywilloccurin2045.[272]

Transhumanism

Youawakeonemorningtofindyourbrainhasanotherlobefunctioning.Invisible,thisauxiliary
lobeanswersyourquestionswithinformationbeyondtherealmofyourownmemory,suggests
plausiblecoursesofaction,andasksquestionsthathelpbringoutrelevantfacts.Youquicklycome
torelyonthenewlobesomuchthatyoustopwonderinghowitworks.Youjustuseit.Thisisthe
dreamofartificialintelligence.

Byte,April1985[273]

RobotdesignerHansMoravec,cyberneticistKevinWarwickandinventorRayKurzweilhavepredictedthat
humansandmachineswillmergeinthefutureintocyborgsthataremorecapableandpowerfulthaneither.[274]
Thisidea,calledtranshumanism,whichhasrootsinAldousHuxleyandRobertEttinger,hasbeenillustratedin
fictionaswell,forexampleinthemangaGhostintheShellandthesciencefictionseriesDune.

Inthe1980sartistHajimeSorayama'sSexyRobotsserieswerepaintedandpublishedinJapandepictingthe
actualorganichumanformwithlifelikemuscularmetallicskinsandlater"theGynoids"bookfollowedthat
wasusedbyorinfluencedmoviemakersincludingGeorgeLucasandothercreatives.Sorayamanever
consideredtheseorganicrobotstoberealpartofnaturebutalwaysunnaturalproductofthehumanmind,a
fantasyexistinginthemindevenwhenrealizedinactualform.

EdwardFredkinarguesthat"artificialintelligenceisthenextstageinevolution",anideafirstproposedby
SamuelButler's"DarwinamongtheMachines"(1863),andexpandeduponbyGeorgeDysoninhisbookofthe
samenamein1998.[275]

Infiction
Thoughtcapableartificialbeingshaveappearedasstorytellingdevicessinceantiquity.[14]

Theimplicationsofaconstructedmachineexhibitingartificialintelligencehavebeenapersistentthemein
sciencefictionsincethetwentiethcentury.Earlystoriestypicallyrevolvedaroundintelligentrobots.Theword
"robot"itselfwascoinedbyKarelapekinhis1921playR.U.R.,thetitlestandingfor"Rossum'sUniversal
Robots".Later,theSFwriterIsaacAsimovdevelopedtheThreeLawsofRoboticswhichhesubsequently
exploredinalongseriesofrobotstories.Asimov'slawsareoftenbroughtupduringlaymandiscussionsof
machineethics[276]whilealmostallartificialintelligenceresearchersarefamiliarwithAsimov'slawsthrough
popularculture,theygenerallyconsiderthelawsuselessformanyreasons,oneofwhichistheirambiguity.[277]

ThenovelDoAndroidsDreamofElectricSheep?,byPhilipK.Dick,tellsasciencefictionstoryabout
Androidsandhumansclashinginafuturisticworld.Elementsofartificialintelligenceincludetheempathybox,
moodorgan,andtheandroidsthemselves.Throughoutthenovel,Dickportraystheideathathumansubjectivity
isalteredbytechnologycreatedwithartificialintelligence.[278]

NowadaysAIisfirmlyrootedinpopularcultureintelligentrobotsappearininnumerableworks.HAL,the
murderouscomputerinchargeofthespaceshipin2001:ASpaceOdyssey(1968),isanexampleofthe
common"roboticrampage"archetypeinsciencefictionmovies.TheTerminator(1984)andTheMatrix(1999)
provideadditionalwidelyfamiliarexamples.Incontrast,therareloyalrobotssuchasGortfromTheDaythe
EarthStoodStill(1951)andBishopfromAliens(1986)arelessprominentinpopularculture.[279]

Seealso
https://en.wikipedia.org/wiki/Artificial_intelligence 20/42
8/18/2017 ArtificialintelligenceWikipedia

Abductivereasoning
Casebasedreasoning
Commonsensereasoning
Emergentalgorithm
Evolutionarycomputing
Glossaryofartificialintelligence
Machinelearning
Mathematicaloptimization
Softcomputing
Swarmintelligence

Notes
1.Theintelligentagentparadigm:
Russell&Norvig2003,pp.27,3258,968972
Poole,Mackworth&Goebel1998,pp.721
Luger&Stubblefield2004,pp.235240
Hutter2005,pp.125126
Thedefinitionusedinthisarticle,intermsofgoals,actions,perceptionandenvironment,isdueto
Russell&Norvig(2003).Otherdefinitionsalsoincludeknowledgeandlearningasadditionalcriteria.
2.Russell&Norvig2009,p.2.
3.Hofstadter(1980,p.601)
4.Schank,RogerC.(1991)."Where'stheAI".AImagazine.Vol.12no.4.p.38.
5.Russell&Norvig2009.
6."AlphaGoGoogleDeepMind"(https://deepmind.com/alphago.html).
7.PamelaMcCorduck(2004,pp.424)writesof"theroughshatteringofAIinsubfieldsvision,natural
language,decisiontheory,geneticalgorithms,robotics...andthesewithownsubsubfieldthatwould
hardlyhaveanythingtosaytoeachother."
8.ThislistofintelligenttraitsisbasedonthetopicscoveredbythemajorAItextbooks,including:
Russell&Norvig2003
Luger&Stubblefield2004
Poole,Mackworth&Goebel1998
Nilsson1998
9.Generalintelligence(strongAI)isdiscussedinpopularintroductionstoAI:
Kurzweil1999andKurzweil2005
10.SeetheDartmouthproposal,underPhilosophy,below.
11.ThisisacentralideaofPamelaMcCorduck'sMachinesWhoThink.Shewrites:"Iliketothinkof
artificialintelligenceasthescientificapotheosisofavenerableculturaltradition."(McCorduck2004,
p.34)"ArtificialintelligenceinoneformoranotherisanideathathaspervadedWesternintellectual
history,adreaminurgentneedofbeingrealized."(McCorduck2004,p.xviii)"Ourhistoryisfullof
attemptsnutty,eerie,comical,earnest,legendaryandrealtomakeartificialintelligences,to
reproducewhatistheessentialusbypassingtheordinarymeans.Backandforthbetweenmythand
reality,ourimaginationssupplyingwhatourworkshopscouldn't,wehaveengagedforalongtimeinthis
oddformofselfreproduction."(McCorduck2004,p.3)ShetracesthedesirebacktoitsHellenisticroots
andcallsittheurgeto"forgetheGods."(McCorduck2004,pp.340400)
12."StephenHawkingbelievesAIcouldbemankind'slastaccomplishment"(https://betanews.com/2016/10/
21/artificialintelligencestephenhawking/).BetaNews.21October2016.
13.AIapplicationswidelyusedbehindthescenes:
Russell&Norvig2003,p.28
Kurzweil2005,p.265
NRC1999,pp.216222
14.AIinmyth:
McCorduck2004,pp.45
Russell&Norvig2003,p.939
15.Russell&Norvig2009,p.16.
16.AIinearlysciencefiction.
McCorduck2004,pp.1725
https://en.wikipedia.org/wiki/Artificial_intelligence 21/42
8/18/2017 ArtificialintelligenceWikipedia

17.Formalreasoning:
Berlinski,David(2000).TheAdventoftheAlgorithm.HarcourtBooks.ISBN0156013916.
OCLC46890682(https://www.worldcat.org/oclc/46890682).
18.AI'simmediateprecursors:
McCorduck2004,pp.51107
Crevier1993,pp.2732
Russell&Norvig2003,pp.15,940
Moravec1988,p.3
19.Dartmouthconference:
McCorduck2004,pp.111136
Crevier1993,pp.4749,whowrites"theconferenceisgenerallyrecognizedastheofficial
birthdateofthenewscience."
Russell&Norvig2003,p.17,whocalltheconference"thebirthofartificialintelligence."
NRC1999,pp.200201
20.HegemonyoftheDartmouthconferenceattendees:
Russell&Norvig2003,p.17,whowrite"forthenext20yearsthefieldwouldbedominatedby
thesepeopleandtheirstudents."
McCorduck2004,pp.129130
21.Russell&Norvig2003,p.18.
22."Goldenyears"ofAI(successfulsymbolicreasoningprograms19561973):
McCorduck2004,pp.243252
Crevier1993,pp.52107
Moravec1988,p.9
Russell&Norvig2003,pp.1821
TheprogramsdescribedareArthurSamuel'scheckersprogramfortheIBM701,DanielBobrow's
STUDENT,NewellandSimon'sLogicTheoristandTerryWinograd'sSHRDLU.
23.DARPApoursmoneyintoundirectedpureresearchintoAIduringthe1960s:
McCorduck2004,pp.131
Crevier1993,pp.51,6465
NRC1999,pp.204205
24.AIinEngland:
Howe1994
25.OptimismofearlyAI:
HerbertSimonquote:Simon1965,p.96quotedinCrevier1993,p.109.
MarvinMinskyquote:Minsky1967,p.2quotedinCrevier1993,p.109.
26.Lighthill1973.
27.FirstAIWinter,MansfieldAmendment,Lighthillreport
Crevier1993,pp.115117
Russell&Norvig2003,p.22
NRC1999,pp.212213
Howe1994
28.Expertsystems:
ACM1998,I.2.1
Russell&Norvig2003,pp.2224
Luger&Stubblefield2004,pp.227331
Nilsson1998,chpt.17.4
McCorduck2004,pp.327335,434435
Crevier1993,pp.14562,197203
29.Boomofthe1980s:riseofexpertsystems,FifthGenerationProject,Alvey,MCC,SCI:
McCorduck2004,pp.426441
Crevier1993,pp.161162,197203,211,240
Russell&Norvig2003,p.24
NRC1999,pp.210211
30.SecondAIwinter:
McCorduck2004,pp.430435
Crevier1993,pp.209210
NRC1999,pp.214216

https://en.wikipedia.org/wiki/Artificial_intelligence 22/42
8/18/2017 ArtificialintelligenceWikipedia

31.Formalmethodsarenowpreferred("Victoryoftheneats"):
Russell&Norvig2003,pp.2526
McCorduck2004,pp.486487
32.McCorduck2004,pp.480483.
33.Deeplearning:
citationinprogress
34.MachinelearningandAI'ssuccessesintheearly21stcentury:
citationinprogress
35.Markoff2011.
36.Administrator."Kinect'sAIbreakthroughexplained"(http://www.iprogrammer.info/news/105artificiali
ntelligence/2176kinectsaibreakthroughexplained.html).iprogrammer.info.
37.Rowinski,Dan(15January2013)."VirtualPersonalAssistants&TheFutureOfYourSmartphone
[Infographic]"(http://readwrite.com/2013/01/15/virtualpersonalassistantsthefutureofyoursmartphon
einfographic).ReadWrite.
38."Artificialintelligence:Google'sAlphaGobeatsGomasterLeeSedol"(http://www.bbc.com/news/techn
ology35785875).BBCNews.12March2016.Retrieved1October2016.
39."AfterWininChina,AlphaGosDesignersExploreNewAI"(https://www.wired.com/2017/05/winchina
alphagosdesignersexplorenewai/).20170527.
40."World'sGoPlayerRatings"(http://www.goratings.org/).May2017.
41."19"(http://sports.sina.com.cn/go/20160802/docifxun
yya3020238.shtml)(inChinese).May2017.
42.Clark,Jack(8December2015)."Why2015WasaBreakthroughYearinArtificialIntelligence"(https://
www.bloomberg.com/news/articles/20151208/why2015wasabreakthroughyearinartificialintellige
nce).BloombergNews.Retrieved23November2016."Afterahalfdecadeofquietbreakthroughsin
artificialintelligence,2015hasbeenalandmarkyear.Computersaresmarterandlearningfasterthan
ever."
43.Sandewall,Erik."TheGoalsofArtificialIntelligenceResearchABriefintroduction".Knowledge
RepresentationFrameworkProjectLinkopingUniversity.N.p.,8August2010.8December2016.
44.Problemsolving,puzzlesolving,gameplayinganddeduction:
Russell&Norvig2003,chpt.39,
Poole,Mackworth&Goebel1998,chpt.2,3,7,9,
Luger&Stubblefield2004,chpt.3,4,6,8,
Nilsson1998,chpt.712
45.Uncertainreasoning:
Russell&Norvig2003,pp.452644,
Poole,Mackworth&Goebel1998,pp.345395,
Luger&Stubblefield2004,pp.333381,
Nilsson1998,chpt.19
46.Intractabilityandefficiencyandthecombinatorialexplosion:
Russell&Norvig2003,pp.9,2122
47.Psychologicalevidenceofsubsymbolicreasoning:
Wason&Shapiro(1966)showedthatpeopledopoorlyoncompletelyabstractproblems,butifthe
problemisrestatedtoallowtheuseofintuitivesocialintelligence,performancedramatically
improves.(SeeWasonselectiontask)
Kahneman,Slovic&Tversky(1982)haveshownthatpeopleareterribleatelementaryproblems
thatinvolveuncertainreasoning.(Seelistofcognitivebiasesforseveralexamples).
Lakoff&Nez(2000)havecontroversiallyarguedthatevenourskillsatmathematicsdependon
knowledgeandskillsthatcomefrom"thebody",i.e.sensorimotorandperceptualskills.(See
WhereMathematicsComesFrom)
48.Sikos,LeslieF.(June2017).DescriptionLogicsinMultimediaReasoning(http://www.springer.com/us/b
ook/9783319540658).Cham:Springer.ISBN9783319540665.doi:10.1007/9783319540665(http
s://doi.org/10.1007%2F9783319540665).
49.Bertini,MDelBimbo,ATorniai,C(2006)."Automaticannotationandsemanticretrievalofvideo
sequencesusingmultimediaontologies".MM06Proceedingsofthe14thACMinternationalconference
onMultimedia.14thACMinternationalconferenceonMultimedia.SantaBarbara:ACM.pp.679682.

https://en.wikipedia.org/wiki/Artificial_intelligence 23/42
8/18/2017 ArtificialintelligenceWikipedia

50.Qualificationproblem:
McCarthy&Hayes1969
Russell&Norvig2003
WhileMcCarthywasprimarilyconcernedwithissuesinthelogicalrepresentationofactions,Russell&
Norvig2003applythetermtothemoregeneralissueofdefaultreasoninginthevastnetworkof
assumptionsunderlyingallourcommonsenseknowledge.
51.Defaultreasoninganddefaultlogic,nonmonotoniclogics,circumscription,closedworldassumption,
abduction(Pooleetal.placesabductionunder"defaultreasoning".Lugeretal.placesthisunder
"uncertainreasoning"):
Russell&Norvig2003,pp.354360,
Poole,Mackworth&Goebel1998,pp.248256,323335,
Luger&Stubblefield2004,pp.335363,
Nilsson1998,~18.3.3
52.Breadthofcommonsenseknowledge:
Russell&Norvig2003,p.21,
Crevier1993,pp.113114,
Moravec1988,p.13,
Lenat&Guha1989(Introduction)
53.Dreyfus&Dreyfus1986.
54.Gladwell2005.
55.Expertknowledgeasembodiedintuition:
Dreyfus&Dreyfus1986(HubertDreyfusisaphilosopherandcriticofAIwhowasamongthe
firsttoarguethatmostusefulhumanknowledgewasencodedsubsymbolically.SeeDreyfus'
critiqueofAI)
Gladwell2005(Gladwell'sBlinkisapopularintroductiontosubsymbolicreasoningand
knowledge.)
Hawkins&Blakeslee2005(Hawkinsarguesthatsubsymbolicknowledgeshouldbetheprimary
focusofAIresearch.)
56.Planning:
ACM1998,~I.2.8,
Russell&Norvig2003,pp.375459,
Poole,Mackworth&Goebel1998,pp.281316,
Luger&Stubblefield2004,pp.314329,
Nilsson1998,chpt.10.12,22
57.Informationvaluetheory:
Russell&Norvig2003,pp.600604
58.Classicalplanning:
Russell&Norvig2003,pp.375430,
Poole,Mackworth&Goebel1998,pp.281315,
Luger&Stubblefield2004,pp.314329,
Nilsson1998,chpt.10.12,22
59.Planningandactinginnondeterministicdomains:conditionalplanning,executionmonitoring,
replanningandcontinuousplanning:
Russell&Norvig2003,pp.430449
60.Multiagentplanningandemergentbehavior:
Russell&Norvig2003,pp.449455
61.AlanTuringdiscussedthecentralityoflearningasearlyas1950,inhisclassicpaper"Computing
MachineryandIntelligence".(Turing1950)In1956,attheoriginalDartmouthAIsummerconference,
RaySolomonoffwroteareportonunsupervisedprobabilisticmachinelearning:"AnInductiveInference
Machine".(Solomonoff1956)
62.ThisisaformofTomMitchell'swidelyquoteddefinitionofmachinelearning:"Acomputerprogramis
settolearnfromanexperienceEwithrespecttosometaskTandsomeperformancemeasurePifits
performanceonTasmeasuredbyPimproveswithexperienceE."

https://en.wikipedia.org/wiki/Artificial_intelligence 24/42
8/18/2017 ArtificialintelligenceWikipedia

63.Learning:
ACM1998,I.2.6,
Russell&Norvig2003,pp.649788,
Poole,Mackworth&Goebel1998,pp.397438,
Luger&Stubblefield2004,pp.385542,
Nilsson1998,chpt.3.3,10.3,17.5,20
64.Reinforcementlearning:
Russell&Norvig2003,pp.763788
Luger&Stubblefield2004,pp.442449
65.Computationallearningtheory:
CITATIONINPROGRESS.
66.Wengetal.2001.
67.Lungarellaetal.2003.
68.Asadaetal.2009.
69.Oudeyer2010.
70.Naturallanguageprocessing:
ACM1998,I.2.7
Russell&Norvig2003,pp.790831
Poole,Mackworth&Goebel1998,pp.91104
Luger&Stubblefield2004,pp.591632
71."Versatilequestionansweringsystems:seeinginsynthesis"(https://www.academia.edu/2475776/Versatil
e_question_answering_systems_seeing_in_synthesis),Mittaletal.,IJIIDS,5(2),119142,2011
72.Applicationsofnaturallanguageprocessing,includinginformationretrieval(i.e.textmining)and
machinetranslation:
Russell&Norvig2003,pp.840857,
Luger&Stubblefield2004,pp.623630
73.Machineperception:
Russell&Norvig2003,pp.537581,863898
Nilsson1998,~chpt.6
74.Computervision:
ACM1998,I.2.10
Russell&Norvig2003,pp.863898
Nilsson1998,chpt.6
75.Speechrecognition:
ACM1998,~I.2.7
Russell&Norvig2003,pp.568578
76.Objectrecognition:
Russell&Norvig2003,pp.885892
77.Robotics:
ACM1998,I.2.9,
Russell&Norvig2003,pp.901942,
Poole,Mackworth&Goebel1998,pp.443460
78.Movingandconfigurationspace:
Russell&Norvig2003,pp.916932
79.Tecuci2012.
80.Roboticmapping(localization,etc):
Russell&Norvig2003,pp.908915
81.Kismet.
82.Thro1993.
83.Edelson1991.
84.Tao&Tan2005.
85.James1884.
86.Picard1995.
87.KleineCosack2006:"TheintroductionofemotiontocomputersciencewasdonebyPickard(sic)who
createdthefieldofaffectivecomputing."
88.Diamond2003:"RosalindPicard,agenialMITprofessor,isthefield'sgodmotherher1997book,
AffectiveComputing,triggeredanexplosionofinterestintheemotionalsideofcomputersandtheir
users."
https://en.wikipedia.org/wiki/Artificial_intelligence 25/42
8/18/2017 ArtificialintelligenceWikipedia

89.Emotionandaffectivecomputing:
Minsky2006
90.Roberts,Jacob(2016)."ThinkingMachines:TheSearchforArtificialIntelligence"(https://www.chemhe
ritage.org/distillations/magazine/thinkingmachinesthesearchforartificialintelligence).Distillations.2
(2):1423.Retrieved17February2017.
91.GeraldEdelman,IgorAleksanderandothershavearguedthatartificialconsciousnessisrequiredfor
strongAI.(Aleksander1995Edelman2007)
92.Artificialbrainarguments:AIrequiresasimulationoftheoperationofthehumanbrain
Russell&Norvig2003,p.957
Crevier1993,pp.271and279
Afewofthepeoplewhomakesomeformoftheargument:
Moravec1988
Kurzweil2005,p.262
Hawkins&Blakeslee2005
Themostextremeformofthisargument(thebrainreplacementscenario)wasputforwardbyClark
Glymourinthemid1970sandwastouchedonbyZenonPylyshynandJohnSearlein1980.
93.NilsNilssonwrites:"Simplyput,thereiswidedisagreementinthefieldaboutwhatAIisallabout"
(Nilsson1983,p.10).
94.Biologicalintelligencevs.intelligenceingeneral:
Russell&Norvig2003,pp.23,whomaketheanalogywithaeronauticalengineering.
McCorduck2004,pp.100101,whowritesthatthereare"twomajorbranchesofartificial
intelligence:oneaimedatproducingintelligentbehaviorregardlessofhowitwasaccomplioshed,
andtheotheraimedatmodelingintelligentprocessesfoundinnature,particularlyhumanones."
Kolata1982,apaperinScience,whichdescribesMcCarthy'sindifferencetobiologicalmodels.
KolataquotesMcCarthyaswriting:"ThisisAI,sowedon'tcareifit'spsychologicallyreal"[1](htt
ps://books.google.com/books?id=PEkqAAAAMAAJ&q=%22we+don't+care+if+it's+psychologica
lly+real%22&dq=%22we+don't+care+if+it's+psychologically+real%22&output=html&pgis=1).
McCarthyrecentlyreiteratedhispositionattheAI@50conferencewherehesaid"Artificial
intelligenceisnot,bydefinition,simulationofhumanintelligence"(Maker2006).
95.Neatsvs.scruffies:
McCorduck2004,pp.421424,486489
Crevier1993,pp.168
Nilsson1983,pp.1011
96.Symbolicvs.subsymbolicAI:
Nilsson(1998,p.7),whousestheterm"subsymbolic".
97.Haugeland1985,p.255.
98.Law1994.
99.Bach2008.
100.Shapiro,StuartC.(1992),"ArtificialIntelligence",inStuartC.Shapiro(ed.),EncyclopediaofArtificial
Intelligence,2ndedition(NewYork:JohnWiley&Sons):5457.4December2016.
101.Haugeland1985,pp.112117
102.ThemostdramaticcaseofsubsymbolicAIbeingpushedintothebackgroundwasthedevastating
critiqueofperceptronsbyMarvinMinskyandSeymourPapertin1969.SeeHistoryofAI,AIwinter,or
FrankRosenblatt.
103.Cognitivesimulation,NewellandSimon,AIatCMU(thencalledCarnegieTech):
McCorduck2004,pp.139179,245250,322323(EPAM)
Crevier1993,pp.145149
104.Soar(history):
McCorduck2004,pp.450451
Crevier1993,pp.258263
105.McCarthyandAIresearchatSAILandSRIInternational:
McCorduck2004,pp.251259
Crevier1993
106.AIresearchatEdinburghandinFrance,birthofProlog:
Crevier1993,pp.193196
Howe1994

https://en.wikipedia.org/wiki/Artificial_intelligence 26/42
8/18/2017 ArtificialintelligenceWikipedia

107.AIatMITunderMarvinMinskyinthe1960s:
McCorduck2004,pp.259305
Crevier1993,pp.83102,163176
Russell&Norvig2003,p.19
108.Cyc:
McCorduck2004,p.489,whocallsit"adeterminedlyscruffyenterprise"
Crevier1993,pp.239243
Russell&Norvig2003,p.363365
Lenat&Guha1989
109.Knowledgerevolution:
McCorduck2004,pp.266276,298300,314,421
Russell&Norvig2003,pp.2223
110.EmbodiedapproachestoAI:
McCorduck2004,pp.454462
Brooks1990
Moravec1988
111.Revivalofconnectionism:
Crevier1993,pp.214215
Russell&Norvig2003,p.25
112.Computationalintelligence
IEEEComputationalIntelligenceSociety(http://www.ieeecis.org/)
113.Hutter2012.
114.Langley2011.
115.Katz2012.
116.Norvig2012.
117.Agentarchitectures,hybridintelligentsystems:
Russell&Norvig(2003,pp.27,932,970972)
Nilsson(1998,chpt.25)
118.Hierarchicalcontrolsystem:
Albus2002
119.Subsumptionarchitecture:
CITATIONINPROGRESS.
120.Searchalgorithms:
Russell&Norvig2003,pp.59189
Poole,Mackworth&Goebel1998,pp.113163
Luger&Stubblefield2004,pp.79164,193219
Nilsson1998,chpt.712
121.Forwardchaining,backwardchaining,Hornclauses,andlogicaldeductionassearch:
Russell&Norvig2003,pp.217225,280294
Poole,Mackworth&Goebel1998,pp.~4652
Luger&Stubblefield2004,pp.6273
Nilsson1998,chpt.4.2,7.2
122.Statespacesearchandplanning:
Russell&Norvig2003,pp.382387
Poole,Mackworth&Goebel1998,pp.298305
Nilsson1998,chpt.10.12
123.Uninformedsearches(breadthfirstsearch,depthfirstsearchandgeneralstatespacesearch):
Russell&Norvig2003,pp.5993
Poole,Mackworth&Goebel1998,pp.113132
Luger&Stubblefield2004,pp.79121
Nilsson1998,chpt.8
124.Heuristicorinformedsearches(e.g.,greedybestfirstandA*):
Russell&Norvig2003,pp.94109,
Poole,Mackworth&Goebel1998,pp.pp.132147,
Luger&Stubblefield2004,pp.133150,
Nilsson1998,chpt.9

https://en.wikipedia.org/wiki/Artificial_intelligence 27/42
8/18/2017 ArtificialintelligenceWikipedia

125.Optimizationsearches:
Russell&Norvig2003,pp.110116,120129
Poole,Mackworth&Goebel1998,pp.56163
Luger&Stubblefield2004,pp.127133
126.Artificiallifeandsocietybasedlearning:
Luger&Stubblefield2004,pp.530541
127.Geneticprogrammingandgeneticalgorithms:
Luger&Stubblefield2004,pp.509530,
Nilsson1998,chpt.4.2,
Holland1975,
Koza1992,
Poli,Langdon&McPhee2008.
128.Logic:
ACM1998,~I.2.3,
Russell&Norvig2003,pp.194310,
Luger&Stubblefield2004,pp.3577,
Nilsson1998,chpt.1316
129.Satplan:
Russell&Norvig2003,pp.402407,
Poole,Mackworth&Goebel1998,pp.300301,
Nilsson1998,chpt.21
130.Explanationbasedlearning,relevancebasedlearning,inductivelogicprogramming,casebased
reasoning:
Russell&Norvig2003,pp.678710,
Poole,Mackworth&Goebel1998,pp.414416,
Luger&Stubblefield2004,pp.~422442,
Nilsson1998,chpt.10.3,17.5
131.Propositionallogic:
Russell&Norvig2003,pp.204233,
Luger&Stubblefield2004,pp.4550
Nilsson1998,chpt.13
132.Firstorderlogicandfeaturessuchasequality:
ACM1998,~I.2.4,
Russell&Norvig2003,pp.240310,
Poole,Mackworth&Goebel1998,pp.268275,
Luger&Stubblefield2004,pp.5062,
Nilsson1998,chpt.15
133.Fuzzylogic:
Russell&Norvig2003,pp.526527
134.Subjectivelogic:
CITATIONINPROGRESS.
135.Representingcategoriesandrelations:Semanticnetworks,descriptionlogics,inheritance(including
framesandscripts):
Russell&Norvig2003,pp.349354,
Poole,Mackworth&Goebel1998,pp.174177,
Luger&Stubblefield2004,pp.248258,
Nilsson1998,chpt.18.3
136.Representingeventsandtime:Situationcalculus,eventcalculus,fluentcalculus(includingsolvingthe
frameproblem):
Russell&Norvig2003,pp.328341,
Poole,Mackworth&Goebel1998,pp.281298,
Nilsson1998,chpt.18.2
137.Causalcalculus:
Poole,Mackworth&Goebel1998,pp.335337
138."TheBeliefCalculusandUncertainReasoning",YenTehHsia
139.Representingknowledgeaboutknowledge:Beliefcalculus,modallogics:
Russell&Norvig2003,pp.341344,
Poole,Mackworth&Goebel1998,pp.275277
https://en.wikipedia.org/wiki/Artificial_intelligence 28/42
8/18/2017 ArtificialintelligenceWikipedia

140.Stochasticmethodsforuncertainreasoning:
ACM1998,~I.2.3,
Russell&Norvig2003,pp.462644,
Poole,Mackworth&Goebel1998,pp.345395,
Luger&Stubblefield2004,pp.165191,333381,
Nilsson1998,chpt.19
141.Bayesiannetworks:
Russell&Norvig2003,pp.492523,
Poole,Mackworth&Goebel1998,pp.361381,
Luger&Stubblefield2004,pp.~182190,363379,
Nilsson1998,chpt.19.34
142.Bayesianinferencealgorithm:
Russell&Norvig2003,pp.504519,
Poole,Mackworth&Goebel1998,pp.361381,
Luger&Stubblefield2004,pp.~363379,
Nilsson1998,chpt.19.4&7
143.Bayesianlearningandtheexpectationmaximizationalgorithm:
Russell&Norvig2003,pp.712724,
Poole,Mackworth&Goebel1998,pp.424433,
Nilsson1998,chpt.20
144.BayesiandecisiontheoryandBayesiandecisionnetworks:
Russell&Norvig2003,pp.597600
145.Stochastictemporalmodels:
Russell&Norvig2003,pp.537581
DynamicBayesiannetworks:
Russell&Norvig2003,pp.551557
HiddenMarkovmodel:
(Russell&Norvig2003,pp.549551)
Kalmanfilters:
Russell&Norvig2003,pp.551557
146.decisiontheoryanddecisionanalysis:
Russell&Norvig2003,pp.584597,
Poole,Mackworth&Goebel1998,pp.381394
147.Markovdecisionprocessesanddynamicdecisionnetworks:
Russell&Norvig2003,pp.613631
148.Gametheoryandmechanismdesign:
Russell&Norvig2003,pp.631643
149.Statisticallearningmethodsandclassifiers:
Russell&Norvig2003,pp.712754,
Luger&Stubblefield2004,pp.453541
150.Neuralnetworksandconnectionism:
Russell&Norvig2003,pp.736748,
Poole,Mackworth&Goebel1998,pp.408414,
Luger&Stubblefield2004,pp.453505,
Nilsson1998,chpt.3
151.kernelmethodssuchasthesupportvectormachine:
Russell&Norvig2003,pp.749752
152.Knearestneighboralgorithm:
Russell&Norvig2003,pp.733736
153.Gaussianmixturemodel:
Russell&Norvig2003,pp.725727
154.NaiveBayesclassifier:
Russell&Norvig2003,pp.718
155.Decisiontree:
Russell&Norvig2003,pp.653664,
Poole,Mackworth&Goebel1998,pp.403408,
Luger&Stubblefield2004,pp.408417

https://en.wikipedia.org/wiki/Artificial_intelligence 29/42
8/18/2017 ArtificialintelligenceWikipedia

156.Classifierperformance:
vanderWalt&Bernard2006
157.Feedforwardneuralnetworks,perceptronsandradialbasisnetworks:
Russell&Norvig2003,pp.739748,758
Luger&Stubblefield2004,pp.458467
158.Competitivelearning,Hebbiancoincidencelearning,Hopfieldnetworksandattractornetworks:
Luger&Stubblefield2004,pp.474505
159.SeppoLinnainmaa(1970).Therepresentationofthecumulativeroundingerrorofanalgorithmasa
Taylorexpansionofthelocalroundingerrors.Master'sThesis(inFinnish),Univ.Helsinki,67.
160.Griewank,Andreas(2012).WhoInventedtheReverseModeofDifferentiation?.OptimizationStories,
DocumentaMatematica,ExtraVolumeISMP(2012),389400.
161.PaulWerbos,"BeyondRegression:NewToolsforPredictionandAnalysisintheBehavioralSciences",
PhDthesis,HarvardUniversity,1974.
162.PaulWerbos(1982).Applicationsofadvancesinnonlinearsensitivityanalysis.InSystemmodelingand
optimization(pp.762770).SpringerBerlinHeidelberg.Online(http://werbos.com/Neural/SensitivityIFI
PSeptember1981.pdf)
163.Backpropagation:
Russell&Norvig2003,pp.744748,
Luger&Stubblefield2004,pp.467474,
Nilsson1998,chpt.3.3
164.Hierarchicaltemporalmemory:
Hawkins&Blakeslee2005
165.IanGoodfellow,YoshuaBengio,andAaronCourville(2016).DeepLearning.MITPress.Online(http://
www.deeplearningbook.org)
166.Hinton,G.Deng,L.Yu,D.Dahl,G.Mohamed,A.Jaitly,N.Senior,A.Vanhoucke,V.Nguyen,P.
Sainath,T.Kingsbury,B.(2012)."DeepNeuralNetworksforAcousticModelinginSpeechRecognition
Thesharedviewsoffourresearchgroups".IEEESignalProcessingMagazine.29(6):8297.
doi:10.1109/msp.2012.2205597(https://doi.org/10.1109%2Fmsp.2012.2205597).
167.Schmidhuber,J.(2015)."DeepLearninginNeuralNetworks:AnOverview".NeuralNetworks.61:85
117.arXiv:1404.7828(https://arxiv.org/abs/1404.7828) .doi:10.1016/j.neunet.2014.09.003(https://doi.
org/10.1016%2Fj.neunet.2014.09.003).
168.Schmidhuber,Jrgen(2015)."DeepLearning"(http://www.scholarpedia.org/article/Deep_Learning).
Scholarpedia.10(11):32832.doi:10.4249/scholarpedia.32832(https://doi.org/10.4249%2Fscholarpedia.
32832).
169.RinaDechter(1986).Learningwhilesearchinginconstraintsatisfactionproblems.Universityof
California,ComputerScienceDepartment,CognitiveSystemsLaboratory.Online(https://www.researchg
ate.net/publication/221605378_Learning_While_Searching_in_ConstraintSatisfactionProblems)
170.IgorAizenberg,NaumN.Aizenberg,JoosP.L.Vandewalle(2000).MultiValuedandUniversalBinary
Neurons:Theory,LearningandApplications.SpringerScience&BusinessMedia.
171.Ivakhnenko,Alexey(1965).CyberneticPredictingDevices.Kiev:NaukovaDumka.
172.Ivakhnenko,Alexey(1971)."Polynomialtheoryofcomplexsystems".IEEETransactionsonSystems,
ManandCybernetics(4):364378.
173.Hinton2007.
174.Research,AI(23October2015)."DeepNeuralNetworksforAcousticModelinginSpeechRecognition"
(http://airesearch.com/airesearchpapers/deepneuralnetworksforacousticmodelinginspeechrecogni
tion/).airesearch.com.Retrieved23October2015.
175.Fukushima,K.(1980)."Neocognitron:Aselforganizingneuralnetworkmodelforamechanismof
patternrecognitionunaffectedbyshiftinposition".BiologicalCybernetics.36:193202.PMID7370364
(https://www.ncbi.nlm.nih.gov/pubmed/7370364).doi:10.1007/bf00344251(https://doi.org/10.1007%2F
bf00344251).
176.YannLeCun(2016).SlidesonDeepLearningOnline(https://indico.cern.ch/event/510372/)
177."AlphaGoGoogleDeepMind"(https://www.deepmind.com/alphago.html).Retrieved30January2016.
178.Recurrentneuralnetworks,Hopfieldnets:
Russell&Norvig2003,p.758
Luger&Stubblefield2004,pp.474505
179.P.J.Werbos.Generalizationofbackpropagationwithapplicationtoarecurrentgasmarketmodel"
NeuralNetworks1,1988.

https://en.wikipedia.org/wiki/Artificial_intelligence 30/42
8/18/2017 ArtificialintelligenceWikipedia

180.A.J.RobinsonandF.Fallside.Theutilitydrivendynamicerrorpropagationnetwork.TechnicalReport
CUED/FINFENG/TR.1,CambridgeUniversityEngineeringDepartment,1987.
181.R.J.WilliamsandD.Zipser.Gradientbasedlearningalgorithmsforrecurrentnetworksandtheir
computationalcomplexity.InBackpropagation:Theory,ArchitecturesandApplications.Hillsdale,NJ:
Erlbaum,1994.
182.SeppHochreiter(1991),UntersuchungenzudynamischenneuronalenNetzen(http://people.idsia.ch/~jue
rgen/SeppHochreiter1991ThesisAdvisorSchmidhuber.pdf),Diplomathesis.Institutf.Informatik,
TechnischeUniv.Munich.Advisor:J.Schmidhuber.
183.Schmidhuber,J.(1992)."Learningcomplex,extendedsequencesusingtheprincipleofhistory
compression".NeuralComputation.4:234242.doi:10.1162/neco.1992.4.2.234(https://doi.org/10.116
2%2Fneco.1992.4.2.234).
184.Hochreiter,SeppandSchmidhuber,JrgenLongShortTermMemory,NeuralComputation,9(8):1735
1780,1997
185.AlexGraves,SantiagoFernandez,FaustinoGomez,andJrgenSchmidhuber(2006).Connectionist
temporalclassification:Labellingunsegmentedsequencedatawithrecurrentneuralnets.Proceedingsof
ICML06,pp.369376.
186.AwniHannun,CarlCase,JaredCasper,BryanCatanzaro,GregDiamos,ErichElsen,RyanPrenger,
SanjeevSatheesh,ShubhoSengupta,AdamCoates,AndrewNg(2014).DeepSpeech:Scalingupendto
endspeechrecognition.arXiv:1412.5567
187.HasimSakandAndrewSeniorandFrancoiseBeaufays(2014).LongShortTermMemoryrecurrent
neuralnetworkarchitecturesforlargescaleacousticmodeling.ProceedingsofInterspeech2014.
188.XiangangLi,XihongWu(2015).ConstructingLongShortTermMemorybasedDeepRecurrentNeural
NetworksforLargeVocabularySpeechRecognitionarXiv:1410.4281
189.HaimSak,AndrewSenior,KanishkaRao,FranoiseBeaufaysandJohanSchalkwyk(September2015):
Googlevoicesearch:fasterandmoreaccurate.(http://googleresearch.blogspot.ch/2015/09/googlevoice
searchfasterandmore.html)
190.IlyaSutskever,OriolVinyals,andQuocV.Le(2014).SequencetoSequenceLearningwithNeural
Networks.arXiv:1409.3215
191.RafalJozefowicz,OriolVinyals,MikeSchuster,NoamShazeer,YonghuiWu(2016).Exploringthe
LimitsofLanguageModeling.arXiv:1602.02410
192.DanGillick,CliffBrunk,OriolVinyals,AmarnagSubramanya(2015).MultilingualLanguage
ProcessingFromBytes.arXiv:1512.00103
193.OriolVinyals,AlexanderToshev,SamyBengio,andDumitruErhan(2015).ShowandTell:ANeural
ImageCaptionGenerator.arXiv:1411.4555
194.Controltheory:
ACM1998,~I.2.8,
Russell&Norvig2003,pp.926932
195.Lisp:
Luger&Stubblefield2004,pp.723821
Crevier1993,pp.5962,
Russell&Norvig2003,p.18
196.Prolog:
Poole,Mackworth&Goebel1998,pp.477491,
Luger&Stubblefield2004,pp.641676,575581
197.TheTuringtest:
Turing'soriginalpublication:
Turing1950
Historicalinfluenceandphilosophicalimplications:
Haugeland1985,pp.69
Crevier1993,p.24
McCorduck2004,pp.7071
Russell&Norvig2003,pp.23and948
198.SubjectmatterexpertTuringtest:
CITATIONINPROGRESS.
199.GameAI:
CITATIONINPROGRESS.

https://en.wikipedia.org/wiki/Artificial_intelligence 31/42
8/18/2017 ArtificialintelligenceWikipedia

200.Mathematicaldefinitionsofintelligence:
HernandezOrallo2000
Dowe&Hajek1997
HernandezOrallo&Dowe2010
201.O'Brien&Marakas2011.
202.Russell&Norvig2009,p.1.
203.CNN2006.
204.N.AletrasD.TsarapatsanisD.PreotiucPietroV.Lampos(2016)."Predictingjudicialdecisionsofthe
EuropeanCourtofHumanRights:aNaturalLanguageProcessingperspective"(https://peerj.com/article
s/cs93/).PeerJComputerScience.
205."TheEconomistExplains:Whyfirmsarepilingintoartificialintelligence"(https://www.economist.com/
blogs/economistexplains/2016/04/economistexplains).TheEconomist.31March2016.Retrieved
19May2016.
206.Lohr,Steve(28February2016)."ThePromiseofArtificialIntelligenceUnfoldsinSmallSteps"(https://
www.nytimes.com/2016/02/29/technology/thepromiseofartificialintelligenceunfoldsinsmallsteps.ht
ml?ref=technology).TheNewYorkTimes.Retrieved29February2016.
207.Wakefield,Jane(June15,2016)."Socialmedia'outstripsTV'asnewssourceforyoungpeople"(http://w
ww.bbc.co.uk/news/uk36528256).BBCNews.
208.Smith,Mark(July22,2016)."Soyouthinkyouchosetoreadthisarticle?"(http://www.bbc.co.uk/news/b
usiness36837824).BBCNews.
209.DinaBass(September20,2016)."MicrosoftDevelopsAItoHelpCancerDoctorsFindtheRight
Treatments"(https://www.bloomberg.com/news/articles/20160920/microsoftdevelopsaitohelpcance
rdoctorsfindtherighttreatments).Bloomberg.
210.Gallagher,James(20170126)."Artificialintelligence'asgoodascancerdoctors' "(http://www.bbc.co.u
k/news/health38717928).BBCNews.Retrieved20170126.
211.[2](https://worldwide.espacenet.com/textdoc?DB=EPODOC&IDX=),Langen,PaulineA.JeffreyS.
Katz&GayleDempsey,"Remotemonitoringofhighriskpatientsusingartificialintelligence"
212.Senthilingam,Meera(May12,2016)."AreAutonomousRobotsYournextSurgeons?"(http://www.cnn.c
om/2016/05/12/health/robotsurgeonboweloperation/).CNN.CableNewsNetwork.Retrieved
December4,2016.
213."33CorporationsWorkingOnAutonomousVehicles".CBInsights.N.p.,11August2016.12November
2016.
214.West,DarrellM."Movingforward:SelfdrivingvehiclesinChina,Europe,Japan,Korea,andtheUnited
States".CenterforTechnologyInnovationatBrookings.N.p.,September2016.12November2016.
215.McFarland,Matt."Google'sartificialintelligencebreakthroughmayhaveahugeimpactonselfdriving
carsandmuchmore".TheWashingtonPost25February2015.InfotracNewsstand.24October2016
216."Programmingsafetyintoselfdrivingcars".NationalScienceFoundation.N.p.,2February2015.24
October2016.
217.O'Neill,,Eleanor(31July2016)."Accounting,automationandAI"(https://www.icas.com/catodaynew
s/howaccountancyandfinanceareusingartificialintelligence).www.icas.com.Retrieved20161118.
218.RobotsBeatHumansinTradingBattle.(http://news.bbc.co.uk/2/hi/business/1481339.stm)BBC.com
(August8,2001)
219."CTOCorner:ArtificialIntelligenceUseinFinancialServicesFinancialServicesRoundtable"(http://fs
roundtable.org/ctocornerartificialintelligenceuseinfinancialservices/).FinancialServices
Roundtable.20150402.Retrieved20161118.
220."WhyAIresearcherslikevideogames"(http://www.economist.com/news/scienceandtechnology/21721
890gameshelpthemunderstandrealitywhyairesearchersvideogames).TheEconomist.
221.Brooks1991.
222."HackingRoomba"(http://hackingroomba.com/).hackingroomba.com.
223."Aselforganizingthousandrobotswarm"(https://www.seas.harvard.edu/news/2014/08/selforganizingt
housandrobotswarm).www.seas.harvard.edu.14August2014.
224."WatchAnAutonomousRobotSwarmForm2DStarfishes"(https://creators.vice.com/en_us/article/watc
hanautonomousrobotswarmform2dstarfishes).Creators.
225.Rainie,LeeJannaerson(8February2017)."Theme2:Goodthingslieahead"(http://www.pewinternet.
org/2017/02/08/theme2goodthingslieahead/).
226.Lynley,Matthew."SoundHoundraises$75MtobringitsvoiceenabledAIeverywhere"(http://social.tec
hcrunch.com/2017/01/31/soundhoundraises75mtobringitsvoiceenabledaieverywhere/).
227."PartnershiponArtificialIntelligencetoBenefitPeopleandSociety".N.p.,n.d.24October2016.
https://en.wikipedia.org/wiki/Artificial_intelligence 32/42
8/18/2017 ArtificialintelligenceWikipedia

228.Fiegerman,Seth."Facebook,Google,AmazonCreateGrouptoEaseAIConcerns".CNNMoney.n.d.4
December2016.
229.Dartmouthproposal:
McCarthyetal.1955(theoriginalproposal)
Crevier1993,p.49(historicalsignificance)
230.Thephysicalsymbolsystemshypothesis:
Newell&Simon1976,p.116
McCorduck2004,p.153
Russell&Norvig2003,p.18
231.Dreyfuscriticizedthenecessaryconditionofthephysicalsymbolsystemhypothesis,whichhecalledthe
"psychologicalassumption":"Themindcanbeviewedasadeviceoperatingonbitsofinformation
accordingtoformalrules."(Dreyfus1992,p.156)
232.Dreyfus'critiqueofartificialintelligence:
Dreyfus1972,Dreyfus&Dreyfus1986
Crevier1993,pp.120132
McCorduck2004,pp.211239
Russell&Norvig2003,pp.950952,
233.Gdel1951:inthislecture,KurtGdelusestheincompletenesstheoremtoarriveatthefollowing
disjunction:(a)thehumanmindisnotaconsistentfinitemachine,or(b)thereexistDiophantine
equationsforwhichitcannotdecidewhethersolutionsexist.Gdelfinds(b)implausible,andthusseems
tohavebelievedthehumanmindwasnotequivalenttoafinitemachine,i.e.,itspowerexceededthatof
anyfinitemachine.Herecognizedthatthiswasonlyaconjecture,sinceonecouldneverdisprove(b).Yet
heconsideredthedisjunctiveconclusiontobea"certainfact".
234.TheMathematicalObjection:
Russell&Norvig2003,p.949
McCorduck2004,pp.448449
MakingtheMathematicalObjection:
Lucas1961
Penrose1989
RefutingMathematicalObjection:
Turing1950under"(2)TheMathematicalObjection"
Hofstadter1979
Background:
Gdel1931,Church1936,Kleene1935,Turing1937
235.GrahamOppy(20January2015)."Gdel'sIncompletenessTheorems"(http://plato.stanford.edu/entries/g
oedelincompleteness/#GdeArgAgaMec).StanfordEncyclopediaofPhilosophy.Retrieved27April2016.
"TheseGdelianantimechanistargumentsare,however,problematic,andthereiswideconsensusthat
theyfail."
236.StuartJ.RussellPeterNorvig(2010)."26.1.2:PhilosophicalFoundations/WeakAI:CanMachinesAct
Intelligently?/Themathematicalobjection".ArtificialIntelligence:AModernApproach(3rded.).Upper
SaddleRiver,NJ:PrenticeHall.ISBN0136042597."...evenifwegrantthatcomputershave
limitationsonwhattheycanprove,thereisnoevidencethathumansareimmunefromthoselimitations."
237.MarkColyvan.Anintroductiontothephilosophyofmathematics.CambridgeUniversityPress,2012.
From2.2.2,'PhilosophicalsignificanceofGdel'sincompletenessresults':"Theacceptedwisdom(with
whichIconcur)isthattheLucasPenroseargumentsfail."
238.Russel,Stuart.,DanielDewey,andMaxTegmark.ResearchPrioritiesforRobustandBeneficial
ArtificialIntelligence.AIMagazine36:4(2015).8December2016.
239."StephenHawkingwarnsartificialintelligencecouldendmankind"(http://www.bbc.com/news/technolo
gy30290540).BBCNews.Retrieved20151030.
240.Holley,Peter(28January2015)."BillGatesondangersofartificialintelligence:'Idon'tunderstandwhy
somepeoplearenotconcerned' "(https://www.washingtonpost.com/news/theswitch/wp/2015/01/28/bill
gatesondangersofartificialintelligencedontunderstandwhysomepeoplearenotconcerned/).The
WashingtonPost.ISSN01908286(https://www.worldcat.org/issn/01908286).Retrieved20151030.
241.Gibbs,Samuel."ElonMusk:artificialintelligenceisourbiggestexistentialthreat"(https://www.theguard
ian.com/technology/2014/oct/27/elonmuskartificialintelligenceaibiggestexistentialthreat).the
Guardian.Retrieved20151030.

https://en.wikipedia.org/wiki/Artificial_intelligence 33/42
8/18/2017 ArtificialintelligenceWikipedia

242.Post,Washington."TechtitanslikeElonMuskarespending$1billiontosaveyoufromterminators"(htt
p://www.chicagotribune.com/bluesky/technology/cttechtitansagainstterminators20151214
story.html).
243.Mller,VincentC.Bostrom,Nick(2014)."FutureProgressinArtificialIntelligence:APollAmong
Experts"(http://www.sophia.de/pdf/2014_PTAI_polls.pdf)(PDF).AIMatters.1(1):911.
doi:10.1145/2639475.2639478(https://doi.org/10.1145%2F2639475.2639478).
244."Isartificialintelligencereallyanexistentialthreattohumanity?"(http://thebulletin.org/artificialintellige
ncereallyexistentialthreathumanity8577).BulletinoftheAtomicScientists.Retrieved20151030.
245."Thecaseagainstkillerrobots,fromaguyactuallyworkingonartificialintelligence"(http://fusion.net/st
ory/54583/thecaseagainstkillerrobotsfromaguyactuallybuildingai/).Fusion.net.Retrieved
20160131.
246."Willartificialintelligencedestroyhumanity?Hereare5reasonsnottoworry."(https://www.vox.com/20
14/8/22/6043635/5reasonsweshouldntworryaboutsuperintelligentcomputerstaking).Vox.
Retrieved20151030.
247."ThemysteriousartificialintelligencecompanyElonMuskinvestedinisdevelopinggamechanging
smartcomputers"(http://www.techinsider.io/mysteriousartificialintelligencecompanyelonmuskinves
tment201510).TechInsider.Retrieved20151030.
248.Clark,Jack."MuskBackedGroupProbesRisksBehindArtificialIntelligence"(https://www.bloomberg.
com/news/articles/20150701/muskbackedgroupprobesrisksbehindartificialintelligence).
Bloomberg.com.Retrieved20151030.
249."ElonMuskIsDonating$10MOfHisOwnMoneyToArtificialIntelligenceResearch"(http://www.fast
company.com/3041007/fastfeed/elonmuskisdonating10mofhisownmoneytoartificialintelligence
research).FastCompany.Retrieved20151030.
250."StephenHawking,ElonMusk,andBillGatesWarnAboutArtificialIntelligence"(http://observer.com/2
015/08/stephenhawkingelonmuskandbillgateswarnaboutartificialintelligence/).Observer.
Retrieved20151030.
251.Intheearly1970s,KennethColbypresentedaversionofWeizenbaum'sELIZAknownasDOCTOR
whichhepromotedasaserioustherapeutictool.(Crevier1993,pp.132144)
252.JosephWeizenbaum'scritiqueofAI:
Weizenbaum1976
Crevier1993,pp.132144
McCorduck2004,pp.356373
Russell&Norvig2003,p.961
Weizenbaum(theAIresearcherwhodevelopedthefirstchatterbotprogram,ELIZA)arguedin1976that
themisuseofartificialintelligencehasthepotentialtodevaluehumanlife.
253.Ford,2009&Thelightsinthetunnel.
254."MachineLearning:Ajobkiller?"(http://econfuture.wordpress.com/2011/04/14/machinelearningajob
killer/).econfutureRobots,AIandUnemploymentFutureEconomicsandTechnology.
255.AIcoulddecreasethedemandforhumanlabor:
Russell&Norvig2003,pp.960961
Ford,Martin(2009).TheLightsintheTunnel:Automation,AcceleratingTechnologyandthe
EconomyoftheFuture(http://www.thelightsinthetunnel.com).AcculantPublishing.ISBN9781
448659814.
256.WendellWallach(2010).MoralMachines,OxfordUniversityPress.
257.Wallach,pp3754.
258.Wallach,pp5573.
259.Wallach,Introductionchapter.
260.MichaelAndersonandSusanLeighAnderson(2011),MachineEthics,CambridgeUniversityPress.
261."MachineEthics"(https://web.archive.org/web/20141129044821/http://www.aaai.org/Library/Symposia/
Fall/fs0506).aaai.org.Archivedfromtheoriginal(http://www.aaai.org/Library/Symposia/Fall/fs0506)
on29November2014.
262.Rubin,Charles(Spring2003)."ArtificialIntelligenceandHumanNature|`TheNewAtlantis"(https://we
b.archive.org/web/20120611115223/http://www.thenewatlantis.com/publications/artificialintelligencean
dhumannature).1:88100.Archivedfromtheoriginal(http://www.thenewatlantis.com/publications/art
ificialintelligenceandhumannature)on11June2012.
263.Rawlinson,Kevin."Microsoft'sBillGatesinsistsAIisathreat"(http://www.bbc.co.uk/news/31047780).
BBCNews.Retrieved30January2015.

https://en.wikipedia.org/wiki/Artificial_intelligence 34/42
8/18/2017 ArtificialintelligenceWikipedia

264.Brooks,Rodney(10November2014)."artificialintelligenceisatool,notathreat"(https://web.archive.o
rg/web/20141112130954/http://www.rethinkrobotics.com/artificialintelligencetoolthreat/).Archived
fromtheoriginal(http://www.rethinkrobotics.com/artificialintelligencetoolthreat/)on12November
2014.
265.Horst,Steven,(2005)"TheComputationalTheoryofMind"(http://plato.stanford.edu/entries/computatio
nalmind/)inTheStanfordEncyclopediaofPhilosophy
266.ThisversionisfromSearle(1999),andisalsoquotedinDennett1991,p.435.Searle'soriginal
formulationwas"Theappropriatelyprogrammedcomputerreallyisamind,inthesensethatcomputers
giventherightprogramscanbeliterallysaidtounderstandandhaveothercognitivestates."(Searle
1980,p.1).StrongAIisdefinedsimilarlybyRussell&Norvig(2003,p.947):"Theassertionthat
machinescouldpossiblyactintelligently(or,perhapsbetter,actasiftheywereintelligent)iscalledthe
'weakAI'hypothesisbyphilosophers,andtheassertionthatmachinesthatdosoareactuallythinking(as
opposedtosimulatingthinking)iscalledthe'strongAI'hypothesis."
267.Searle'sChineseroomargument:
Searle1980.Searle'soriginalpresentationofthethoughtexperiment.
Searle1999.
Discussion:
Russell&Norvig2003,pp.958960
McCorduck2004,pp.443445
Crevier1993,pp.269271
268.Robotrights:
Russell&Norvig2003,p.964
BBCNews2006
Prematurityof:
Henderson2007
Infiction:
McCorduck(2004,p.19025)discussesFrankensteinandidentifiesthekeyethicalissuesas
scientifichubrisandthesufferingofthemonster,i.e.robotrights.
269.Evans,Woody(2015)."PosthumanRights:DimensionsofTranshumanWorlds"(http://revistas.ucm.es/in
dex.php/TEKN/article/view/49072/46310).Teknokultura.UniversidadComplutense,Madrid.Retrieved
December5,2016.
270.maschafilm."Content:Plug&PrayFilmArtificialIntelligenceRobots"(http://www.plugandprayfil
m.de/en/content.html).plugandprayfilm.de.
271.Omohundro,Steve(2008).TheNatureofSelfImprovingArticialIntelligence.presentedanddistributed
atthe2007SingularitySummit,SanFrancisco,CA.
272.Technologicalsingularity:
Vinge1993
Kurzweil2005
Russell&Norvig2003,p.963
273.Lemmons,Phil(April1985)."ArtificialIntelligence"(https://archive.org/stream/bytemagazine198504/
1985_04_BYTE_1004_Artificial_Intelligence#page/n125/mode/2up).BYTE.p.125.Retrieved
14February2015.
274.Transhumanism:
Moravec1988
Kurzweil2005
Russell&Norvig2003,p.963
275.AIasevolution:
EdwardFredkinisquotedinMcCorduck(2004,p.401).
Butler1863
Dyson1998
276.Anderson,SusanLeigh."Asimovs"threelawsofrobotics"andmachinemetaethics."AI&Society22.4
(2008):477493.
277.McCauley,Lee(2007)."AIarmageddonandthethreelawsofrobotics".EthicsandInformation
Technology.9(2):153164.doi:10.1007/s1067600791382(https://doi.org/10.1007%2Fs1067600791
382).
278.Galvan,Jill(19970101)."EnteringthePosthumanCollectiveinPhilipK.Dick's"DoAndroidsDream
ofElectricSheep?" ".ScienceFictionStudies.24(3):413429.JSTOR4240644(https://www.jstor.org/st
able/4240644).
https://en.wikipedia.org/wiki/Artificial_intelligence 35/42
8/18/2017 ArtificialintelligenceWikipedia

279.Buttazzo,G.(July2001)."Artificialconsciousness:Utopiaorrealpossibility?"(http://ieeexplore.ieee.or
g/document/933500/?reload=true).Computer(IEEE).34(7):2430.doi:10.1109/2.933500(https://doi.or
g/10.1109%2F2.933500).Retrieved29December2016.

Citeerror:Alistdefinedreferencenamed"Knowledge_representation"isnotusedinthecontent(see
thehelppage).
Citeerror:Alistdefinedreferencenamed"Knowledge_engineering"isnotusedinthecontent(seethe
helppage).
Citeerror:Alistdefinedreferencenamed"Ontology"isnotusedinthecontent(seethehelppage).

References
AItextbooks

Hutter,Marcus(2005).UniversalArtificialIntelligence.Berlin:Springer.ISBN9783540221395.
Luger,GeorgeStubblefield,William(2004).ArtificialIntelligence:StructuresandStrategiesfor
ComplexProblemSolving(http://www.cs.unm.edu/~luger/aifinal/tocfull.html)(5thed.).
Benjamin/Cummings.ISBN0805347801.
Neapolitan,RichardJiang,Xia(2012).ContemporaryArtificialIntelligence(http://www.crcpress.com/p
roduct/isbn/9781439844694).Chapman&Hall/CRC.ISBN9781439844694.
Nilsson,Nils(1998).ArtificialIntelligence:ANewSynthesis.MorganKaufmann.ISBN978155860
4674.
Russell,StuartJ.Norvig,Peter(2003),ArtificialIntelligence:AModernApproach(http://aima.cs.berkel
ey.edu/)(2nded.),UpperSaddleRiver,NewJersey:PrenticeHall,ISBN0137903952.
Russell,StuartJ.Norvig,Peter(2009).ArtificialIntelligence:AModernApproach(3rded.).Upper
SaddleRiver,NewJersey:PrenticeHall.ISBN0136042597..
Poole,DavidMackworth,AlanGoebel,Randy(1998).ComputationalIntelligence:ALogical
Approach(http://www.cs.ubc.ca/spider/poole/ci.html).NewYork:OxfordUniversityPress.ISBN019
5102703.
Winston,PatrickHenry(1984).ArtificialIntelligence.Reading,MA:AddisonWesley.ISBN0201
082594.
Rich,Elaine(1983).ArtificialIntelligence.McGrawHill.ISBN0070522618.
Bundy,Alan(1980).ArtificialIntelligence:AnIntroductoryCourse(2nded.).EdinburghUniversity
Press.ISBN085224410X.

HistoryofAI

Crevier,Daniel(1993),AI:TheTumultuousSearchforArtificialIntelligence,NewYork,NY:
BasicBooks,ISBN0465029973.
McCorduck,Pamela(2004),MachinesWhoThink(http://www.pamelamc.com/html/machines_who_thin
k.html)(2nded.),Natick,MA:A.K.Peters,Ltd.,ISBN1568812051.
Newquist,HP(1994).TheBrainMakers:Genius,Ego,AndGreedInTheQuestForMachinesThat
Think.NewYork:Macmillan/SAMS.ISBN0672304120.
Nilsson,Nils(2009).TheQuestforArtificialIntelligence:AHistoryofIdeasandAchievements.New
York:CambridgeUniversityPress.ISBN9780521122931.

Othersources

Asada,M.Hosoda,K.Kuniyoshi,Y.Ishiguro,H.Inui,T.Yoshikawa,Y.Ogino,M.Yoshida,C.
(2009)."Cognitivedevelopmentalrobotics:asurvey"(https://web.archive.org/web/20131004222242/htt
p://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=4895715&url=http%3A%2F%2Fieeexplore.ieee.or
g%2Fiel5%2F4563672%2F5038478%2F04895715.pdf%3Farnumber%3D4895715)(PDF).IEEE
TransactionsonAutonomousMentalDevelopment.1(1):1234.doi:10.1109/tamd.2009.2021702(http
s://doi.org/10.1109%2Ftamd.2009.2021702).Archivedfromtheoriginal(http://ieeexplore.ieee.org/xpl/lo
https://en.wikipedia.org/wiki/Artificial_intelligence 36/42
8/18/2017 ArtificialintelligenceWikipedia

gin.jsp?tp=&arnumber=4895715&url=http%3A%2F%2Fieeexplore.ieee.org%2Fiel5%2F4563672%2F50
38478%2F04895715.pdf%3Farnumber%3D4895715)(PDF)on4October2013.
"ACMComputingClassificationSystem:Artificialintelligence"(http://www.acm.org/class/1998/I.2.htm
l).ACM.1998.Retrieved30August2007.
Goodman,Joanna(2016).RobotsinLaw:HowArtificialIntelligenceisTransformingLegalServices(htt
ps://www.arkgroup.com/product/robotslawhowartificialintelligencetransforminglegalservices#.W
CChnforLDc)(1sted.).ArkGroup.ISBN9781783582648.
Albus,J.S.(2002)."4D/RCS:AReferenceModelArchitectureforIntelligentUnmannedGround
Vehicles"(https://web.archive.org/web/20040725051856/http://www.isd.mel.nist.gov/documents/albus/4
DRCS.pdf)(PDF).InGerhart,G.Gunderson,R.Shoemaker,C.ProceedingsoftheSPIEAeroSense
SessiononUnmannedGroundVehicleTechnology.3693.pp.1120.Archivedfromtheoriginal(http://w
ww.isd.mel.nist.gov/documents/albus/4DRCS.pdf)(PDF)on25July2004.
Aleksander,Igor(1995).ArtificialNeuroconsciousness:AnUpdate(https://web.archive.org/web/199703
02014628/http://www.ee.ic.ac.uk/research/neural/publications/iwann.html).IWANN.Archivedfromthe
original(http://www.ee.ic.ac.uk/research/neural/publications/iwann.html)on2March1997.BibTex(htt
p://dblp.unitrier.de/rec/bibtex/conf/iwann/Aleksander95)Archived(https://web.archive.org/web/199703
02014628/http://www.ee.ic.ac.uk/research/neural/publications/iwann.html)2March1997attheWayback
Machine..
Bach,Joscha(2008)."SevenPrinciplesofSyntheticIntelligence"(https://books.google.com/books?id=a
_ZR81Z25z0C&pg=PA63).InWang,PeiGoertzel,BenFranklin,Stan.ArtificialGeneralIntelligence,
2008:ProceedingsoftheFirstAGIConference.IOSPress.pp.6374.ISBN9781586038335.
"Robotscoulddemandlegalrights"(http://news.bbc.co.uk/2/hi/technology/6200005.stm).BBCNews.21
December2006.Retrieved3February2011.
Brooks,Rodney(1990)."ElephantsDon'tPlayChess"(http://people.csail.mit.edu/brooks/papers/elephan
ts.pdf)(PDF).RoboticsandAutonomousSystems.6:315.doi:10.1016/S09218890(05)800259(https://
doi.org/10.1016%2FS09218890%2805%29800259).Archived(https://web.archive.org/web/200708090
20912/http://people.csail.mit.edu/brooks/papers/elephants.pdf)(PDF)fromtheoriginalon9August
2007.
Brooks,R.A.(1991)."Howtobuildcompletecreaturesratherthanisolatedcognitivesimulators".In
VanLehn,K.ArchitecturesforIntelligence.Hillsdale,NJ:LawrenceErlbaumAssociates.pp.225239.
Buchanan,BruceG.(2005)."A(Very)BriefHistoryofArtificialIntelligence"(https://web.archive.org/w
eb/20070926023314/http://www.aaai.org/AITopics/assets/PDF/AIMag2604016.pdf)(PDF).AI
Magazine:5360.Archivedfromtheoriginal(http://www.aaai.org/AITopics/assets/PDF/AIMag260401
6.pdf)(PDF)on26September2007.
Butler,Samuel(13June1863)."DarwinamongtheMachines"(http://www.nzetc.org/tm/scholarly/teiBu
tFirt1g1t1g1t4body.html).LetterstotheEditor.ThePress.Christchurch,NewZealand.Retrieved
16October2014viaVictoriaUniversityofWellington.
"AIsettoexceedhumanbrainpower"(http://www.cnn.com/2006/TECH/science/07/24/ai.bostrom/).
CNN.26July2006.Archived(https://web.archive.org/web/20080219001624/http://www.cnn.com/2006/
TECH/science/07/24/ai.bostrom/)fromtheoriginalon19February2008.
Dennett,Daniel(1991).ConsciousnessExplained.ThePenguinPress.ISBN0713990376.
Diamond,David(December2003)."TheLoveMachineBuildingcomputersthatcare"(https://www.wir
ed.com/wired/archive/11.12/love.html).Wired.Archived(https://web.archive.org/web/20080518185630/
http://www.wired.com/wired/archive/11.12/love.html)fromtheoriginalon18May2008.
Dowe,D.L.Hajek,A.R.(1997)."AcomputationalextensiontotheTuringTest"(http://www.csse.mon
ash.edu.au/publications/1997/trcs97322abs.html).Proceedingsofthe4thConferenceofthe
AustralasianCognitiveScienceSociety.
Dreyfus,Hubert(1972).WhatComputersCan'tDo.NewYork:MITPress.ISBN0060110821.
Dreyfus,HubertDreyfus,Stuart(1986).MindoverMachine:ThePowerofHumanIntuitionand
ExpertiseintheEraoftheComputer.Oxford,UK:Blackwell.ISBN0029080606.
Dreyfus,Hubert(1992).WhatComputersStillCan'tDo.NewYork:MITPress.ISBN0262540673.
Dyson,George(1998).DarwinamongtheMachines.AllanLaneScience.ISBN0738200301.
Edelman,Gerald(23November2007)."GeraldEdelmanNeuralDarwinismandBrainbasedDevices"
(https://web.archive.org/web/20091008184132/http://lis.epfl.ch/resources/podcast/2007/11/geraldedelma
nneuraldarwinismand.html).TalkingRobots.Archivedfromtheoriginal(http://lis.epfl.ch/resources/po
dcast/2007/11/geraldedelmanneuraldarwinismand.html)on8October2009.
Edelson,Edward(1991).TheNervousSystem.NewYork:ChelseaHouse.ISBN9780791004647.

https://en.wikipedia.org/wiki/Artificial_intelligence 37/42
8/18/2017 ArtificialintelligenceWikipedia

Fearn,Nicholas(2007).TheLatestAnswerstotheOldestQuestions:APhilosophicalAdventurewiththe
World'sGreatestThinkers.NewYork:GrovePress.ISBN0802118399.
Gladwell,Malcolm(2005).Blink.NewYork:Little,BrownandCo.ISBN0316172324.
Gdel,Kurt(1951).Somebasictheoremsonthefoundationsofmathematicsandtheirimplications.
GibbsLecture.In
Feferman,Solomon,ed.(1995).KurtGdel:CollectedWorks,Vol.III:UnpublishedEssaysandLectures.
OxfordUniversityPress.pp.30423.ISBN9780195147223.
Haugeland,John(1985).ArtificialIntelligence:TheVeryIdea.Cambridge,Mass.:MITPress.ISBN0
262081539.
Hawkins,JeffBlakeslee,Sandra(2005).OnIntelligence.NewYork,NY:OwlBooks.ISBN08050
78533.
Henderson,Mark(24April2007)."Humanrightsforrobots?We'regettingcarriedaway"(http://www.th
etimes.co.uk/tto/technology/article1966391.ece).TheTimesOnline.London.
HernandezOrallo,Jose(2000)."BeyondtheTuringTest".JournalofLogic,LanguageandInformation.
9(4):447466.doi:10.1023/A:1008367325700(https://doi.org/10.1023%2FA%3A1008367325700).
HernandezOrallo,J.Dowe,D.L.(2010)."MeasuringUniversalIntelligence:TowardsanAnytime
IntelligenceTest".ArtificialIntelligenceJournal.174(18):15081539.doi:10.1016/j.artint.2010.09.006
(https://doi.org/10.1016%2Fj.artint.2010.09.006).
Hinton,G.E.(2007)."Learningmultiplelayersofrepresentation".TrendsinCognitiveSciences.11:
428434.doi:10.1016/j.tics.2007.09.004(https://doi.org/10.1016%2Fj.tics.2007.09.004).
Hofstadter,Douglas(1979).Gdel,Escher,Bach:anEternalGoldenBraid.NewYork,NY:Vintage
Books.ISBN0394745027.
Holland,JohnH.(1975).AdaptationinNaturalandArtificialSystems.UniversityofMichiganPress.
ISBN0262581116.
Howe,J.(November1994)."ArtificialIntelligenceatEdinburghUniversity:aPerspective"(http://www.i
nf.ed.ac.uk/about/AIhistory.html).Retrieved30August2007.
Hutter,M.(2012)."OneDecadeofUniversalArtificialIntelligence".TheoreticalFoundationsof
ArtificialGeneralIntelligence.AtlantisThinkingMachines.4.ISBN9789491216619.
doi:10.2991/9789491216626_5(https://doi.org/10.2991%2F9789491216626_5).
James,William(1884)."WhatisEmotion".Mind.9:188205.doi:10.1093/mind/osIX.34.188(https://d
oi.org/10.1093%2Fmind%2FosIX.34.188).CitedbyTao&Tan2005.
Kahneman,DanielSlovic,D.Tversky,Amos(1982).Judgmentunderuncertainty:Heuristicsand
biases.NewYork:CambridgeUniversityPress.ISBN0521284147.
Katz,Yarden(1November2012)."NoamChomskyonWhereArtificialIntelligenceWentWrong"(http
s://www.theatlantic.com/technology/archive/2012/11/noamchomskyonwhereartificialintelligencewe
ntwrong/261637/?single_page=true).TheAtlantic.Retrieved26October2014.
"Kismet"(http://www.ai.mit.edu/projects/humanoidroboticsgroup/kismet/kismet.html).MITArtificial
IntelligenceLaboratory,HumanoidRoboticsGroup.Retrieved25October2014.
Koza,JohnR.(1992).GeneticProgramming(OntheProgrammingofComputersbyMeansofNatural
Selection).MITPress.ISBN0262111705.
KleineCosack,Christian(October2006)."RecognitionandSimulationofEmotions"(https://web.archiv
e.org/web/20080528135730/http://ls12www.cs.tudortmund.de/~fink/lectures/SS06/humanrobotinterac
tion/EmotionRecognitionAndSimulation.pdf)(PDF).Archivedfromtheoriginal(http://ls12www.cs.tu
dortmund.de//~fink/lectures/SS06/humanrobotinteraction/EmotionRecognitionAndSimulation.pdf)
(PDF)on28May2008.
Kolata,G.(1982)."Howcancomputersgetcommonsense?".Science.217(4566):12371238.
PMID17837639(https://www.ncbi.nlm.nih.gov/pubmed/17837639).doi:10.1126/science.217.4566.1237
(https://doi.org/10.1126%2Fscience.217.4566.1237).
Kumar,GulshanKumar,Krishan(2012)."TheUseofArtificialIntelligenceBasedEnsemblesfor
IntrusionDetection:AReview"(http://www.hindawi.com/journals/acisc/2012/850160/).Applied
ComputationalIntelligenceandSoftComputing.2012:120.doi:10.1155/2012/850160(https://doi.org/1
0.1155%2F2012%2F850160).
Kurzweil,Ray(1999).TheAgeofSpiritualMachines.PenguinBooks.ISBN0670882178.
Kurzweil,Ray(2005).TheSingularityisNear.PenguinBooks.ISBN0670033847.
Lakoff,GeorgeNez,RafaelE.(2000).WhereMathematicsComesFrom:HowtheEmbodiedMind
BringsMathematicsintoBeing.BasicBooks.ISBN0465037712.
Langley,Pat(2011)."Thechangingscienceofmachinelearning".MachineLearning.82(3):275279.
doi:10.1007/s109940115242y(https://doi.org/10.1007%2Fs109940115242y).
https://en.wikipedia.org/wiki/Artificial_intelligence 38/42
8/18/2017 ArtificialintelligenceWikipedia

Law,Diane(June1994).Searle,SubsymbolicFunctionalismandSyntheticIntelligence(Technical
report).UniversityofTexasatAustin.p.AI94222.CiteSeerX10.1.1.38.8384(https://citeseerx.ist.psu.ed
u/viewdoc/summary?doi=10.1.1.38.8384) .
Legg,ShaneHutter,Marcus(15June2007).ACollectionofDefinitionsofIntelligence(Technical
report).IDSIA.arXiv:0706.3639(https://arxiv.org/abs/0706.3639) .0707.
Lenat,DouglasGuha,R.V.(1989).BuildingLargeKnowledgeBasedSystems.AddisonWesley.
ISBN0201517523.
Lighthill,James(1973)."ArtificialIntelligence:AGeneralSurvey".ArtificialIntelligence:apaper
symposium.ScienceResearchCouncil.
Lucas,John(1961)."Minds,MachinesandGdel".InAnderson,A.R.MindsandMachines(http://users.
ox.ac.uk/~jrlucas/Godel/mmg.html).Archived(https://web.archive.org/web/20070819165214/http://user
s.ox.ac.uk/~jrlucas/Godel/mmg.html)fromtheoriginalon19August2007.Retrieved30August2007.
Lungarella,M.Metta,G.Pfeifer,R.Sandini,G.(2003)."Developmentalrobotics:asurvey".
ConnectionScience.15:151190.CiteSeerX10.1.1.83.7615(https://citeseerx.ist.psu.edu/viewdoc/summ
ary?doi=10.1.1.83.7615) .doi:10.1080/09540090310001655110(https://doi.org/10.1080%2F09540090
310001655110).
Maker,MegHouston(2006)."AI@50:AIPast,Present,Future"(https://web.archive.org/web/200701032
22615/http://www.engagingexperience.com/2006/07/ai50_ai_past_pr.html).DartmouthCollege.
Archivedfromtheoriginal(http://www.engagingexperience.com/2006/07/ai50_ai_past_pr.html)on3
January2007.Retrieved16October2008.
Markoff,John(16February2011)."ComputerWinson'Jeopardy!':Trivial,It'sNot"(https://www.nytime
s.com/2011/02/17/science/17jeopardywatson.html).TheNewYorkTimes.Retrieved25October2014.
McCarthy,JohnMinsky,MarvinRochester,NathanShannon,Claude(1955)."AProposalforthe
DartmouthSummerResearchProjectonArtificialIntelligence"(http://wwwformal.stanford.edu/jmc/his
tory/dartmouth/dartmouth.html).Archived(https://web.archive.org/web/20070826230310/http://wwwfo
rmal.stanford.edu/jmc/history/dartmouth/dartmouth.html)fromtheoriginalon26August2007.
Retrieved30August2007..
McCarthy,JohnHayes,P.J.(1969)."Somephilosophicalproblemsfromthestandpointofartificial
intelligence"(http://wwwformal.stanford.edu/jmc/mcchay69.html).MachineIntelligence.4:463502.
Archived(https://web.archive.org/web/20070810233856/http://wwwformal.stanford.edu/jmc/mcchay69.
html)fromtheoriginalon10August2007.Retrieved30August2007.
McCarthy,John(12November2007)."WhatIsArtificialIntelligence?"(http://wwwformal.stanford.ed
u/jmc/whatisai/whatisai.html).
Minsky,Marvin(1967).Computation:FiniteandInfiniteMachines.EnglewoodCliffs,N.J.:Prentice
Hall.ISBN0131654497.
Minsky,Marvin(2006).TheEmotionMachine.NewYork,NY:Simon&Schusterl.ISBN074327663
9.
Moravec,Hans(1988).MindChildren.HarvardUniversityPress.ISBN0674576160.
Norvig,Peter(25June2012)."OnChomskyandtheTwoCulturesofStatisticalLearning"(http://norvig.
com/chomsky.html).PeterNorvig.Archived(https://web.archive.org/web/20141019223259/http://norvi
g.com/chomsky.html)fromtheoriginalon19October2014.
NRC(UnitedStatesNationalResearchCouncil)(1999)."DevelopmentsinArtificialIntelligence".
FundingaRevolution:GovernmentSupportforComputingResearch.NationalAcademyPress.
Needham,Joseph(1986).ScienceandCivilizationinChina:Volume2.CavesBooksLtd.
Newell,AllenSimon,H.A.(1976)."ComputerScienceasEmpiricalInquiry:SymbolsandSearch"(htt
ps://web.archive.org/web/20081007162810/http://www.rci.rutgers.edu/~cfs/472_html/AI_SEARCH/PSS/
PSSH4.html).CommunicationsoftheACM.19(3):113126.doi:10.1145/360018.360022(https://doi.or
g/10.1145%2F360018.360022).Archivedfromtheoriginal(http://www.rci.rutgers.edu/~cfs/472_html/AI
_SEARCH/PSS/PSSH4.html)on7October2008..
Nilsson,Nils(1983)."ArtificialIntelligencePreparesfor2001"(http://ai.stanford.edu/~nilsson/OnlinePu
bsNils/General%20Essays/AIMag0404002.pdf)(PDF).AIMagazine.1(1).PresidentialAddresstothe
AssociationfortheAdvancementofArtificialIntelligence.
O'Brien,JamesMarakas,George(2011).ManagementInformationSystems(10thed.).McGraw
Hill/Irwin.ISBN9780073376813.
O'Connor,KathleenMalone(1994)."Thealchemicalcreationoflife(takwin)andotherconceptsof
GenesisinmedievalIslam"(http://repository.upenn.edu/dissertations/AAI9503804).Universityof
Pennsylvania.

https://en.wikipedia.org/wiki/Artificial_intelligence 39/42
8/18/2017 ArtificialintelligenceWikipedia

Oudeyer,PY.(2010)."Ontheimpactofroboticsinbehavioralandcognitivesciences:frominsect
navigationtohumancognitivedevelopment"(http://www.pyoudeyer.com/IEEETAMDOudeyer10.pdf)
(PDF).IEEETransactionsonAutonomousMentalDevelopment.2(1):216.
doi:10.1109/tamd.2009.2039057(https://doi.org/10.1109%2Ftamd.2009.2039057).
Penrose,Roger(1989).TheEmperor'sNewMind:ConcerningComputer,MindsandTheLawsof
Physics.OxfordUniversityPress.ISBN0198519737.
Picard,Rosalind(1995).AffectiveComputing(http://affect.media.mit.edu/pdfs/95.picard.pdf)(PDF)
(Technicalreport).MIT.321.Laysummary(http://vismod.media.mit.edu/pub/techreports/TR321ABS
TRACT.html)Abstract.
Poli,R.Langdon,W.B.McPhee,N.F.(2008).AFieldGuidetoGeneticProgramming(http://www.gp
fieldguide.org.uk/).Lulu.com.ISBN9781409200734viagpfieldguide.org.uk.
Rajani,Sandeep(2011)."ArtificialIntelligenceManorMachine"(https://web.archive.org/web/201301
18095844/http://www.csjournals.com/IJITKM/PDF%2041/35.Sandeep%20Rajani.pdf)(PDF).
InternationalJournalofInformationTechnologyandKnowledgeManagement.4(1):173176.Archived
fromtheoriginal(http://www.csjournals.com/IJITKM/PDF%2041/35.Sandeep%20Rajani.pdf)(PDF)on
18January2013.
Searle,John(1980)."Minds,BrainsandPrograms"(https://web.archive.org/web/20100118120034/http://
www.bbsonline.org/documents/a/00/00/04/84/bbs0000048400/bbs.searle2.html).BehavioralandBrain
Sciences.3(3):417457.doi:10.1017/S0140525X00005756(https://doi.org/10.1017%2FS0140525X000
05756).Archivedfromtheoriginal(http://www.bbsonline.org/documents/a/00/00/04/84/bbs000004840
0/bbs.searle2.html)on18January2010.
Searle,John(1999).Mind,languageandsociety.NewYork,NY:BasicBooks.ISBN0465045219.
OCLC231867665(https://www.worldcat.org/oclc/231867665).
Shapiro,StuartC.(1992)."ArtificialIntelligence".InShapiro,StuartC.EncyclopediaofArtificial
Intelligence(http://www.cse.buffalo.edu/~shapiro/Papers/ai.pdf)(PDF)(2nded.).NewYork:JohnWiley.
pp.5457.ISBN0471503061.
Simon,H.A.(1965).TheShapeofAutomationforMenandManagement.NewYork:Harper&Row.
Skillings,Jonathan(3July2006)."GettingMachinestoThinkLikeUs"(http://news.cnet.com/Gettingm
achinestothinklikeus/200811394_36090207.html).cnet.Retrieved3February2011.
Solomonoff,Ray(1956).AnInductiveInferenceMachine(http://world.std.com/~rjs/indinf56.pdf)(PDF).
DartmouthSummerResearchConferenceonArtificialIntelligenceviastd.com,pdfscannedcopyof
theoriginal.Laterpublishedas
Solomonoff,Ray(1957)."AnInductiveInferenceMachine".IREConventionRecord.Sectionon
InformationTheory,part2.pp.5662.
Tao,JianhuaTan,Tieniu(2005).AffectiveComputingandIntelligentInteraction.AffectiveComputing:
AReview.LNCS3784.Springer.pp.981995.doi:10.1007/11573548(https://doi.org/10.1007%2F11573
548).
Tecuci,Gheorghe(MarchApril2012)."ArtificialIntelligence".WileyInterdisciplinaryReviews:
ComputationalStatistics.Wiley.4(2):168180.doi:10.1002/wics.200(https://doi.org/10.1002%2Fwics.
200).
Thro,Ellen(1993).Robotics:TheMarriageofComputersandMachines.NewYork:FactsonFile.
ISBN9780816026289.
Turing,Alan(October1950),"ComputingMachineryandIntelligence"(http://loebner.net/Prizef/Turing
Article.html),Mind,LIX(236):433460,ISSN00264423(https://www.worldcat.org/issn/00264423),
doi:10.1093/mind/LIX.236.433(https://doi.org/10.1093%2Fmind%2FLIX.236.433),retrieved
20080818.
vanderWalt,ChristiaanBernard,Etienne(2006)."Datacharacteristicsthatdetermineclassifier
performance"(https://web.archive.org/web/20090325194051/http://www.patternrecognition.co.za/public
ations/cvdwalt_data_characteristics_classifiers.pdf)(PDF).Archivedfromtheoriginal(http://www.patter
nrecognition.co.za/publications/cvdwalt_data_characteristics_classifiers.pdf)(PDF)on25March2009.
Retrieved5August2009.
Vinge,Vernor(1993)."TheComingTechnologicalSingularity:HowtoSurviveinthePostHumanEra"
(http://wwwrohan.sdsu.edu/faculty/vinge/misc/singularity.html).
Wason,P.C.Shapiro,D.(1966)."Reasoning".InFoss,B.M.Newhorizonsinpsychology.
Harmondsworth:Penguin.
Weizenbaum,Joseph(1976).ComputerPowerandHumanReason.SanFrancisco:W.H.Freeman&
Company.ISBN0716704641.

https://en.wikipedia.org/wiki/Artificial_intelligence 40/42
8/18/2017 ArtificialintelligenceWikipedia

Weng,J.McClellandPentland,A.Sporns,O.Stockman,I.Sur,M.Thelen,E.(2001)."Autonomous
mentaldevelopmentbyrobotsandanimals"(http://www.cse.msu.edu/dl/SciencePaper.pdf)(PDF).
Science.291:599600.doi:10.1126/science.291.5504.599(https://doi.org/10.1126%2Fscience.291.5504.
599)viamsu.edu.
"ApplicationsofAI"(http://wwwformal.stanford.edu/jmc/whatisai/node3.html).www
formal.stanford.edu.Retrieved20160925.

Furtherreading
TechCastArticleSeries,JohnSagi,"FramingConsciousness"(https://web.archive.org/web/20150713184
128/https://www.techcastglobal.com/documents/10193/34869/ConsciousnessSagifinalversion)
Boden,Margaret,MindAsMachine,OxfordUniversityPress,2006
Gopnik,Alison,"MakingAIMoreHuman:Artificialintelligencehasstagedarevivalbystartingto
incorporatewhatweknowabouthowchildrenlearn",ScientificAmerican,vol.316,no.6(June2017),
pp.6065.
Johnston,John(2008)TheAllureofMachinicLife:Cybernetics,ArtificialLife,andtheNewAI,MIT
Press
Marcus,Gary,"AmIHuman?:Researchersneednewwaystodistinguishartificialintelligencefromthe
naturalkind",ScientificAmerican,vol.316,no.3(March2017),pp.5863.Multipletestsofartificial
intelligenceefficacyareneededbecause,"justasthereisnosingletestofathleticprowess,therecannot
beoneultimatetestofintelligence."Onesuchtest,a"ConstructionChallenge",wouldtestperception
andphysicalaction"twoimportantelementsofintelligentbehaviorthatwereentirelyabsentfromthe
originalTuringtest."Anotherproposalhasbeentogivemachinesthesamestandardizedtestsofscience
andotherdisciplinesthatschoolchildrentake.Asofarinsuperablestumblingblocktoartificial
intelligenceisanincapacityforreliabledisambiguation."[V]irtuallyeverysentence[thatpeople
generate]isambiguous,ofteninmultipleways."Aprominentexampleisknownasthe"pronoun
disambiguationproblem":amachinehasnowayofdeterminingtowhomorwhatapronounina
sentencesuchas"he","she"or"it"refers.
Myers,CourtneyBoyded.(2009)."TheAIReport"(http://www.forbes.com/2009/06/22/singularityrobo
tscomputersopinionscontributorsartificialintelligence09_land.html).ForbesJune2009
Raphael,Bertram(1976).TheThinkingComputer.W.H.FreemanandCompany.ISBN0716707233.
Serenko,Alexander(2010)."ThedevelopmentofanAIjournalrankingbasedontherevealedpreference
approach"(http://www.aserenko.com/papers/JOI_Serenko_AI_Journal_Ranking_Published.pdf)(PDF).
JournalofInformetrics.4(4):447459.doi:10.1016/j.joi.2010.04.001(https://doi.org/10.1016%2Fj.joi.2
010.04.001).
Serenko,AlexanderMichaelDohan(2011)."Comparingtheexpertsurveyandcitationimpactjournal
rankingmethods:ExamplefromthefieldofArtificialIntelligence"(http://www.aserenko.com/papers/JO
I_AI_Journal_Ranking_Serenko.pdf)(PDF).JournalofInformetrics.5(4):629649.
doi:10.1016/j.joi.2011.06.002(https://doi.org/10.1016%2Fj.joi.2011.06.002).
Sun,R.&Bookman,L.(eds.),ComputationalArchitectures:IntegratingNeuralandSymbolicProcesses.
KluwerAcademicPublishers,Needham,MA.1994.
TomSimonite(29December2014)."2014inComputing:BreakthroughsinArtificialIntelligence"(htt
p://www.technologyreview.com/news/533686/2014incomputingbreakthroughsinartificialintelligenc
e/).MITTechnologyReview.

Externallinks
WhatIsAI?(http://wwwformal.stanford.edu/jmc/whatisai/whatisai.html)Anintroductiontoartificial
intelligencebyJohnMcCarthyacofounderofthefield,andthepersonwhocoinedtheterm.
TheHandbookofArtificialIntelligenceVolumebyAvronBarrandEdwardA.Feigenbaum(Stanford
University)(https://archive.org/details/handbookofartific01barr/)
"ArtificialIntelligence"(http://www.iep.utm.edu/artinte).InternetEncyclopediaofPhilosophy.
Thomason,Richmond."LogicandArtificialIntelligence"(https://plato.stanford.edu/entries/logicai/).
StanfordEncyclopediaofPhilosophy.
AI(https://dmoztools.net/Computers/Artificial_Intelligence/)atDMOZ
https://en.wikipedia.org/wiki/Artificial_intelligence 41/42
8/18/2017 ArtificialintelligenceWikipedia

AITopics(http://aitopics.org/)Alargedirectoryoflinksandotherresourcesmaintainedbythe
AssociationfortheAdvancementofArtificialIntelligence,theleadingorganizationofacademicAI
researchers.

Retrievedfrom"https://en.wikipedia.org/w/index.php?title=Artificial_intelligence&oldid=795946258"

Thispagewaslasteditedon17August2017,at14:22.
TextisavailableundertheCreativeCommonsAttributionShareAlikeLicenseadditionaltermsmay
apply.Byusingthissite,youagreetotheTermsofUseandPrivacyPolicy.Wikipediaisaregistered
trademarkoftheWikimediaFoundation,Inc.,anonprofitorganization.

https://en.wikipedia.org/wiki/Artificial_intelligence 42/42