资源描述:
《专题16 科普类说明文-三学年高考(2015-2017)英语试题(附解析)》由会员上传分享,免费在线阅读,更多相关内容在教育资源-天天文库。
1、韩老师编辑2015-2017年高考英语分项解析精编版专题16科普类说明文2017高考题科普知识类[2017·北京卷]DHollywood'stheorythatmachineswithevil(邪恶的)mindswilldrivearmiesofkillerrobotsisjustsilly.Therealproblemrelatestothepossibilitythatartificialintelligence(AI)maybecomeextremelygoodatachievingsomethingotherthan
2、whatwereallywant.In1960awellknownmathematicianNorbertWiener,whofoundedthefieldofcybernetics(控制论),putitthisway:“Ifweuse,toachieveourpurposes,amechanicalagencywithwhoseoperationwecannoteffectivelyinterfere(干预),wehadbetterbequitesurethatthepurposeputintothemachineisth
3、epurposewhichwereallydesire.”Amachinewithaspecificpurposehasanotherquality,onethatweusuallyassociatewithlivingthings:awishtopreserveitsownexistence.Forthemachine,thisqualityisnotinborn,norisitsomethingintroducedbyhumans;itisalogicalconsequenceofthesimplefactthatthe
4、machinecannotachieveitsoriginalpurposeifitisdead.Soifwesendoutarobotwiththesingleinstructionoffetchingcoffee,itwillhaveastrongdesiretosecuresuccessbydisablingitsownoffswitchorevenkillinganyonewhomightinterferewithitstask.Ifwearenotcareful,then,wecouldfaceakindofglo
5、balchessmatchagainstverydetermined,superintelligentmachineswhoseobjectivesconflictwithourown,withtherealworldasthechessboard.Thepossibilityofenteringintoandlosingsuchamatchshouldconcentratethemindsofcomputerscientists.Someresearchersarguethatwecansealthemachinesins
6、ideakindoffirewall,usingthemtoanswerdifficultquestionsbutneverallowingthemtoaffecttherealworld.Unfortunately,thatplanseemsunlikelytowork:wehaveyettoinventafirewallthatissecureagainstordinaryhumans,letalonesuperintelligentmachines.Solvingthesafetyproblemwellenoughto
7、moveforwardinAIseemstobepossiblebutnot72韩老师编辑easy.Thereareprobablydecadesinwhichtoplanforthearrivalofsuperintelligentmachines.Buttheproblemshouldnotbedismissedoutofhand,asithasbeenbysomeAIresearchers.Somearguethathumansandmachinescancoexistaslongastheyworkinteams—y
8、etthatisnotpossibleunlessmachinessharethegoalsofhumans.Otherssaywecanjust“switchthemoff”asifsuperintelligentmachinesaretoostupidtothinkofthatposs