復(fù)試面試上大專業(yè)英語(yǔ)unit

content(v.)

合理性，理 誹謗適滿足，滿跨學(xué)科4Unit

人類已賦予自己智慧之人的稱號(hào)，因?yàn)槲覀兊哪軐?duì)日常生活和自我感覺(jué)是如此重要

Unit

其本慮施肥，滋不能反駁，推前人的，不免一死慮施肥，滋不能反駁，推前人的，不免一死耗推可理博5AI力求構(gòu)去理解智6

AnotherreasontostudyAIisthatthese ligententitiesareinterestingandusefulintheirownright.AIhasproducedmanysignificantandimpressiveproductsevenatthisearlystageinitsdevelopment.Althoughnoonecanpredictthefutureindetail,itisclearthatcomputerswithhuman-levelin ligence(orbetter)wouldhaveahugeimpactonoureverydaylivesandonthefuturecourseofcivilization.具備 人類智能水平的計(jì)算機(jī)將在的沖擊 7Unit

一個(gè)又慢又小的大腦或電腦如何能感知、理解預(yù)測(cè) 一個(gè)遠(yuǎn)遠(yuǎn)比它大而復(fù)雜的世界

不同于尋求超光速運(yùn)反重力裝 AIaddressesoneoftheultimatepuzzles.Howisitpossibleforaslow,tinybrain,whetherbiologicalorelectronic,toperceive,understand,predict,andmanipulateaworldfarlargerandmorecomplicatedthanitself?Howdowegoaboutmakingsomethingwiththoseproperties?

Thesearehardquestions,butunlikethesearchforfaster-than-lighttraveloranantigravitydevice,theresearcherinAIhassolidevidencethatthequestispossible.1Alltheresearcherhastodoistolookinthemirrortoseeanexampleofan ligent UnitAIisoneofthenewestdisciplines.Itwasformallyinitiatedin1956,whenthenamewascoined,althoughatthatpointworkhadbeenunderwayforaboutfiveyears.Alongwithmoderngenetics,itisregularlycitedasthe“fieldIwouldmostliketobein”byscientistsinotherdisciplines.和現(xiàn)代遺傳學(xué)一起，科學(xué)工作者稱為“最愿意從事的領(lǐng)域”。

AstudentinphysicsmightreasonablyfeelthatallthegoodideashavealreadybeentakenbyGalileo,Newton,Einstein,andtherest,andthatittakesmanyyearsofstudybeforeonecancontributenewideas.AI,ontheotherhand,stillhasopeningsforafull-timeEinstein. Unit從諸如感知和邏輯推理等的通用領(lǐng)域，到下棋、證明數(shù)學(xué)定理、寫詩(shī)AIcurrentlypassesahugevarietyofsubfields,fromgeneral-purposeareassuchasperceptionandlogicalreasoning,tospecifictaskssuchasyingchess,provingmathematicaltheorems,writingpoetry,anddiagnosing

事的智力工作實(shí)現(xiàn)系統(tǒng)化和自動(dòng)化具和語(yǔ)言 Often,scientistsinotherfieldsmovegraduallyintoartificialin ligence,wheretheyfindthetoolsandvocabularytosystematizeandautomatethein tasksonwhichtheyhavebeenworkingalltheirlives.2Similarly,workersinAIcanchoosetoapplytheirmethodstoanyareaofhumanin lectualendeavor.Inthissense,itistrulyauniversalfield.Unit WhatIs UnitWehavenowex inedwhyAIisexciting,butwehavenotsaidwhatitis.Wecouldjustsay,“Well,ithastodowithsmartprograms,solet’sgetonandwritesome.”Butthehistoryofscienceshowsthatitishelpfultoaimattherightgoals.Earlyalchemists,lookingforapotionforeternallifeandamethodtoturnleadintogold,wereprobablyoffonthewrongfoot.早年尋點(diǎn)金術(shù)士恐怕是邁錯(cuò)了

只有改變目標(biāo)，用早期天文學(xué)家星宿和行星運(yùn)動(dòng)那樣的方法尋求能給出人間世界準(zhǔn)確的明確Onlywhentheaimchanged,tothatoffindingexplicittheoriesthatgaveaccuratepredictionsoftheterrestrialworld,inthesamewaythatearlyastronomypredictedtheapparentmotionsofthestarsand nets,couldthescientificmethodemergeandproductivesciencetake Unit上面的是關(guān)于思維和推理過(guò)程下面的是關(guān)于DefinitionsasgiveniinTable12.1varyalongtwomaainndimensions.Theonesontopareconcernedwiththoughtprocessesandreasoning,whereastheonesonthebottomaddressbehavior.Also,thedefinitionsontheleftmeasuresuccessintermsofhumanperformance,whereastheonesontherightmeasureagainstanidealconceptofin whichwewillcallrationality.Asystemisrationalifitdoestherightthing.

Unit

SystemsthatthinklikeSystemsthatthinkSystemsthatactSystemsSystemsthatthinklikeSystemsthatthinkSystemsthatactSystemsthat人的表…智能概… Unit以人為中心的方法和以理中心的方法之間存在Asonemightexpect,atensionexistsbetweenapproachescenteredaroundhumansandapproachescenteredaroundrationality.Weshouldpointoutthatbydistinguishingbetweenhumanandrationalbehavior,wearenotsuggestingthathumansarenecessarily“irrational”inthesenseof“emotionallyunstable”or“insane”. 我們并不是在“情續(xù)不穩(wěn)”和“精神失常的意義上暗示人必然沒(méi)有理性

UnitOnemerelyneednotethatweoftenmakemistakes;wearenotallchessgrandmasterseventhoughwemayknowalltherulesofchess;andunfortuna y,noteveryonegetsanAontheexam.Ahuman-centeredapproachmustbeanempiricalscience,involvinghypothesisandexperimental Arationalistapproachinvolvesacombinationofmathematicsandengineering.Peopleineachgroupsometimescastaspersionsonworkdoneintheothergroups,butthetruthisthateachdirectionhasyieldedvaluableinsights.Letuslookateachinmoredetail.

Unit ActingHumanly:theTuringTestTheTuringTest,proposedbyAlanTuring,wasdesignedtoprovideasatisfactoryoperationaldefinitionofin Turingdefinedin ligentbehaviorastheabilitytoachievehuman-levelperformanceinallcognitivetasks,sufficienttofoolaninterrogator.Roughlyspeaking,thetestheproposedisthatthecomputershouldbeinterrogatedbyahumanviaaetype,andpassesthetestiftheinterrogatorcannot lifthereisacomputerorahumanattheotherend.UnitDetailsofthetest,andwhetherornotacomputeris ligentifitpassesthetest,willbediscussedlater.Fornow,programmingacomputertopassthetestprovidesplentytoworkon.Thecomputerwouldneedtopossessthefollowing

UnitNaturallanguageprocessingtoenableittocommunicatesuccessfullyinEnglish(orsomeotherhumanKnowledgerepresentationtostoreinformationprovidedorduringtheAutomatedreasoningtousethestoredinformationtoanswerquestionsandtodrawnewconclusions;Machinelearningtoadapttonewcircumstancesandtodetectandextrapolatepatterns.檢測(cè)并向外 Unit避免在詢問(wèn)者和計(jì)算機(jī)之直接物理接Turing’stestdeliberatelyavoideddirectphysicalinteractionbetweentheinterrogatorandthecomputer,becausephysicalsimulationofa isunnecessaryforinligence.However,theso-calledtotalTuringTestincludesa sothattheinterrogatorcantestthesubject’sperceptualabilities,aswellastheopportunityfortheinterrogatortopassphysicalobjects“throughthehatch”.4 詢問(wèn)者就可以對(duì)被測(cè)試者的感覺(jué)能力 試， 時(shí)也包括詢問(wèn)者“通過(guò)艙口”傳遞實(shí)物的可能

UnitTopassthetotalTuringTest,thecomputerwillComputervisiontoperceiveRoboticstomovethemUnitWithinAI,therehasnotbeenabigefforttotrytopasstheTuringtest.TheissueofactinglikeahumancomesupprimarilywhenAIprogramshavetoin ctwithpeople,aswhenanexpertsystemex inshowitcametoitsdiagnosis,oranaturallanguageprocessingsystemhasadialoguewitha自然語(yǔ)言處理系統(tǒng)與用

UnitTheseprogramsmustbehaveaccordingtocertainnormalconventionsofhumanin ctioninordertomakethemselvesunderstood.Theunderlyingrepresentationandreasoninginsuchasystemmayormaynotbebasedonahumanmodel.在這樣一個(gè)系統(tǒng)里基本的表達(dá)和推能是基于人類模型的，也可能不是 Unit17 ThinkingHumanly:theCognitiveModelingApproachIfwearegoingtosaythatagivenprogramthinkslikeahuman,wemusthavesomewayofdetermininghowhumansthink.Weneedtogetinsidetheactualworkingsofhumanminds.Therearetwowaystodothis:throughintrospectiontryingtocatchourownthoughtsastheygobythroughpsychological有兩種方法去做：通過(guò)內(nèi)省，當(dāng)我們的思維現(xiàn)時(shí)抓住它們；或者通過(guò)心理學(xué)實(shí)驗(yàn)

Unit那就證明程序的某些機(jī)理也 以 內(nèi)運(yùn) Oncewehaveasufficientlyprecisetheoryofthemind,itespossibletoexpressthetheoryasacomputerprogram.Iftheprogram’sinput/outputandtimingbehaviormatcheshumanbehavior,thatisevidencethatsomeoftheprogram’smechanismsmayalsobeoperatinginhumans.Forexample,NewellandSimon,whodevelopedGPS,the“GeneralProblemSolver”,werenotcontenttohavetheirprogramcorrectlysolveproblems.發(fā)明“通用解題裝置”的紐厄爾和西蒙并不 Unit

認(rèn)知科學(xué)這一交叉學(xué)科領(lǐng)域?qū)⑷斯ぶ悄茴I(lǐng)域計(jì)算機(jī)模型與心理學(xué)中的實(shí)驗(yàn)技術(shù)結(jié)合起來(lái)，試 Theyweremoreconcernedwithcomparingthetraceofitsreasoningstepstotracesofhumansubjectssolvingthesameproblems.Thisisincontrasttootherresearchersofthesametime,whowereconcernedwithgettingtherightanswersregardlessofhowhumansmightdoit.5

TheinterdisciplinaryfieldofcognitivesciencebringstogethercomputermodelsfromAIandexperimentaltechniquesfrompsychologytotrytoconstructpreciseandtestabletheoriesoftheworkingsofthehumanmind.6 Unit

然而真正的認(rèn)知科學(xué)必定建立際的人

Unit ThinkingRationally:theofThought他著名 法提供了爭(zhēng)論結(jié)構(gòu)的模式， Althoughcognitivescienceisafascinatingfieldinitself,wearenotgoingtobediscussingitallthatmuch.Realcognitivescience,however,isnecessarilybasedonexperimentalinvestigationofactualhumansoranimals.WewillsimplynotethatAIandcognitivesciencecontinuetofertilizeeachother,especiallyintheareasofvision,naturallanguage,and我們只是要說(shuō)明，AI和認(rèn)知科學(xué)繼續(xù)互相促進(jìn)特別是在視覺(jué)領(lǐng)域、自然語(yǔ)言和學(xué)習(xí)方面

定正確的前提就能得出正確的結(jié)TheGreekphilosopherAristotlewasoneofthefirsttoattempttocodify“rightthinking”,thatis,irrefutablereasoningprocesses.Hisfamoussyllogismsprovidedpatternsforargumentstructuresthatalwaysgavecorrectconclusionsgivencorrectpremises.Forexample,“Socratesisaman;allmenaremortal;thereforeSocratesismortal”.Theselawsofthoughtweresupposedto erntheoperationofthemind,andinitiatedthefieldoflogic. 人們假定這些思想法則支配思維活動(dòng)，它們開(kāi)創(chuàng)了邏輯學(xué)領(lǐng)域Unit十九世紀(jì)后期和二十世紀(jì)初形式邏輯的發(fā)展提供了世界上各類事物的精確表示法和他們之間的關(guān)系

Unit

只要提供足夠的時(shí)間 空間，就會(huì)有計(jì)算 程序以邏輯表達(dá)方式描述問(wèn)題，而且只要有解 就能找到這些解 Thedevelopmentofformallogicinthelatenineteenthandearlytwentiethcenturiesprovidedaprecisenotationforstatementsaboutallkindsofthingsintheworldandtherelationsbetweenthem.(Contrastthiswithordinaryarithmeticnotation,whichprovidesmainlyforequalityandinequalitystatementsaboutnumbers.) 與常規(guī)算術(shù)表示法形成對(duì)照，算術(shù)主要關(guān)于數(shù)值相等和不等的表述

By1965,programsexistedthatcould,givenenoughtimeandmemory,takeadescriptionofaprobleminlogicalnotationandfindthesolutiontotheproblem,ifoneexisted.7(Ifthereisnosolution,theprogrammightneverstoplookingforit.)Theso-calledlogicisttraditionwithinartificialin hopestobuildonsuchprogramstocreatein ligentsystems.人工智能領(lǐng)域中所謂的邏輯主義實(shí)踐派構(gòu)造這樣的程序以創(chuàng)建人工智能系統(tǒng) Unit獲取非正式的知識(shí)而用邏輯表示法所需的正式術(shù)述它們并不容易，特別是當(dāng)知識(shí)不是100%確定時(shí)Therearetwomainobstaclestothisapproach.First,itisnoteasytotakeinformalknowledgeandstateitintheformaltermsrequiredbylogicalnotation,particularlywhentheknowledgeislessthan100%certain.Second,thereisabigdifferencebetweenbeingabletosolveaproblem“inprinciple” ngsoinpractice.

Unit即使僅有幾十個(gè)事實(shí)的問(wèn)題就可能耗算機(jī)的資源，除非有關(guān)于首先嘗試哪個(gè)推理驟的某種指Evenproblemswithjustafewdozenfactscanexhaustthecomputationalresourcesofanycomputerunlessithassomeguidanceastowhichreasoningstepstotryfirst.Althoughbothoftheseobstaclesapplytoanyattempttobuildcomputationalreasoningsystems,theyappearedfirstinthelogicisttraditionbecausethepoweroftherepresentationandreasoningsystemsarewell-definedandfairlywellunderstood.雖然這兩個(gè) 推理系統(tǒng)的能力有明確定義并有相當(dāng)好的理解

Unit ActingRationally:RationalAgent理性行事意思是在給 信念的情況下為了達(dá)到其目的而實(shí)行的行Actingrationallymeansactingsoastoachieveone’sgoals,givenone’sbeliefs.Anagentisjustsomethingthatperceivesandacts.(Thismaybeanunusualuseoftheword,butyouwillgetusedtoit.)Inthisapproach,AIisviewedasthestudyandconstructionofrationalagents.AI被看作是研究和構(gòu)建理 UnitInUnitInthe“l(fā)awsoffthought”approachtoAI,thewholeemphasiswasoncorrectinferences.Makingcorrectinferencesissometimespartofbeingarationalagent,becauseonewaytoactrationallyistoreasonlogicallytotheconclusionthatagivenactionwillachieveone’sgoals,andthentoactonthatUnitOntheotherhand,correctinferenceisnotallofrationality,becausethereareoftensituationswherethereisnoprovablycorrectthingtodo,yetsomethingmuststillbedone.TherearealsowaysofactingrationallyTherearealsowaysofactingrationallythatcannotbereasonablysaidtoinvolveinference.Forexample,pullingone’shandoffofahotstoveisareflexactionthatismoresuccessfulthanasloweractiontakenaftercareful 要成功得多。Unit得我們能在各種各樣的情形下達(dá)到好的決策。Allthe“cognitiveskills”neededfortheTuringTestaretheretoallowrationalactions.Thus,weneedtheabilitytorepresentknowledgeandreasonwithitbecausethisenablesustoreachgooddecisionsinawidevarietyofsituations.Weneedto