熱點(diǎn)06DeepSeek人工智能（四大題型）

<1

二泳點(diǎn)解讀

DeepSeek—深度求索：引領(lǐng)智能解決方案的未來

在這個(gè)科技飛速發(fā)展的時(shí)代，人工智能已成為改變世界的關(guān)鍵力量。近日有一款A(yù)I模型正以迅猛之勢在

全球掀起熱潮，它就是DeepSeek。它究竟有著怎樣的獨(dú)特魅力，能在AI領(lǐng)域嶄露頭角？今天，就讓我們

一同揭開DeepSeek的神秘面紗，探尋它如何引領(lǐng)智能解決方案的未來，又將給我們的生活帶來哪些意想

不到的變革。

（建議用時(shí)：60分鐘）

L-----------------W

熱點(diǎn)話題01

閱讀理解

Passage1

Contemporaryinformationretrievalsystemsstandatacrossroads,whereDeepSeek'sneuro-symbolic

architecturefundamentallysubvertsdecades-oldmethodologies.Unliketraditionalmodelsconstrainedbythe

lexical-semanticdichotomy（詞］匚語義二分法）一whereBM25excelsattermfrequencyoptimizationbutfaltersin

contextualinference,orTransformer-basedmodels（BERT,DPR）thatcapturesemanticsubtledifferencesat

prohibitivecomputationalcosts-DeepSeekpioneersadaptivesparse-densefusion（稀疏-稠密結(jié)合）.This

techniquedynamicallyallocatesresourcesthroughagatingmechanism,achieving87.3%precisioninMIT'sTREC

ComplexAnswerRetrievalbenchmarkwhileconsumingmerely23%ofDPR'senergyfootprint.

Themodel'sdynamicsparseactivationmechanism,selectivelyengagingonly12%ofparametersperquery,

contrastssharplywithDSI'smemory-intensivefull-parameteroperations.Stanford'scomparativeanalysisreveals

DeepSeek'slatency-vs-recallParetodominance:11msresponsetimewith94%recallagainstDSI's37ms/96%

tradeoff,acriticaladvantageinreal-timeapplicationslikeemergencyresponsesystems.

However,epistemologicaldebatespersist.WhileDeepSeek'smulti-hopreasoningcapacity（validatedby

92%accuracyonHotpotQA'smulti-documentinferencetasks）surpassesclassicalmodels,itsgradient-disentangled

representationlearningraisesphilosophicalconcerns.The2023ACLmeta-studydocumentsinstanceswherethe

model9scompressedknowledgedistillationinadvertentlyerasedlow-resourcelanguagemorphologicalpatterns,

promptingUNESCO（聯(lián)合國教科文組織）todraftethicalguidelinesforcognitivebiaspropagationinAi-mediated

search.

Commercialadoptionpatternsproveequallytransformative.BloombergreportsthatDeepSeek-poweredfinancial

analyticsplatformsreducedfraudulenttransactionidentificationtimefrom9.2to1.4seconds,outperformingIBM's

WatsonDiscoveryby18.7%inFlscores.Yet,itsblack-boxqueryreformulationprocessremainscontentious-the

EuropeanAIOfficerecentlymandatedtransparencyprotocolsafterDeepSeek'sexplainabilitymetricsscored0.48

ontheLIREframework(vs.0.72forinterpretablesystemslikeElasticLearnedSparseEncoder).Notes:DSI

(DimensionStrategies)isaframeworkrelatedtodimensionstrategiesinthefieldofAIbyMicrosoft.

1.WhattechnologicalbreakthroughenablesDeepSeek'senergyefficiency?

A.Full-parameteroptimization.B.Gradient-disentangledlearning.

C.Adaptivesparse-densefusion.D.Lexical-semanticdichotomy.

2.WhichmetrichighlightsDeepSeek'sreal-timeapplicationsuperiority?

A.92%accuracyonHotpotQAB.11msresponsetimewith94%recall

C.0.48LIREexplainabilityscoreD.87.3%TRECprecisionrate

3.WhatethicalconcernarisesfromDeepSeek'sdesign?

A.Excessiveenergyconsumption.B.Propagationofcognitivebiases.

C.Slowfrauddetectionspeeds.D.Overlytransparentoperations.

4.Howdoestheneuro-symbolicintegrationaddressmodellimitations?

A.Byprioritizingspeedoveraccuracy.

B.Throughcompressedknowledgedistillation.

C.Usingmulti-hopreasoningcapacity.

D.Viadynamicparameteractivation.

Passage2

Withtherapiddevelopmentofartificialintelligence,DeepSeek,aChineseAIcompany,hasemergedasa

pioneerineducationaltechnology.Itslatestmodel,DeepSeek-Rl,combinesadvancedmachinelearningmethodsto

providepersonalizedlearningsolutionsforstudentsworldwide.

UnliketraditionalAImodelsthatrelyonpre-programmedanswers,DeepSeek-Rlusesreinforcement(強(qiáng)化)

learningtosimulatehumanreasoning.Thisallowsittoguidestudentsthroughproblem-solvingstepbystep,much

likeapatienttutor.Forexample,whenastudentstruggleswithamathequation,DeepSeek-Rldoesn?tjustgivethe

answer;itbreaksdownthelogic,identifieserrors,andencouragescriticalthinking.

Themodel'sapplicationsextendbeyondacademics.Inlanguagelearning,itanalyzesstudents5pronunciation

throughAIspeechrecognitionandoffersreal-timefeedback.Forteachers,DeepSeek-Rlcangeneratelessonplans

alignedwithcurriculumstandardsandevenpredictstudents9learningdifficultiesbasedonhistoricaldata.Its

“adaptivetesting^^featurecreatescustomizedquizzesthatadjustdifficultyaccordingtoindividualprogress.

However,challengesremain.Criticsarguethatover-relianceonAImightreducehumaninteractionin

education.DeepSeek'sdevelopersaddressthisbyemphasizingitsroleasa"supplement,notareplacement.AsDr.

Li,aDeepSeekresearcher,stated,t4Ourgoalistofreeteachersfromrepetitivetaskssotheycanfocusoninspiring

creativity.

Lookingahead,DeepSeekaimstointegratevirtualreality(VR)intoitsplatform,allowingstudentsto

explorehistoricaleventsorscientificconceptsinimmersive3Denvironments.WhileethicaldebatesaboutAIin

educationpersist,onethingisclear:toolslikeDeepSeekarereshapinghowwelearn,blendingtechnologywith

humanwisdom.

1.WhatmakesDeepSeek-RldifferentfromtraditionalAImodels?

A.Itusespre-programmedanswers.

B.Itfocusesonmemorizationtechniques.

C.Itreplacesteachersinclassrooms.

D.Itemploysreinforcementlearningforreasoning.

2.Theunderlinedword“adaptive“inParagraph3mostlikelymeans.

A.fixedB.adjustable

C.complicatedD.outdated

3.WhatcanbeinferredaboutDeepSeek'sfutureplans?

A.VRtechnologywillenhanceinteractivelearning.

B.Itwilleliminateallclassroomactivities.

C.Itprioritizesprofitovereducationalvalues.

D.TeacherswilllosetheirjobstoAL

4.Whatisthemainpurposeofthepassage?

A.ToadvertiseDeepSeekproducts.

B.TocriticizetherisksofAIinschools.

C.TodiscussAFsroleintransformingeducation.

D.TocomparedifferentAImodels.

Passage3

DeepSeek:PioneeringtheFutureofIntelligentSolutions

Intheageofrapidtechnologicaldevelopment,DeepSeekhasbecomeapioneerinartificialintelligence(AI)

anddataanalytics.Itwasfoundedtohelpindustriesmakeintelligentdecisions.DeepSeekfocusesoncreating

advancedAIplatformstoboostefficiencyandproductivityinfieldslikefinance,healthcare,education,and

logistics.

DeepSeekusesadvancedtechnologiessuchasmachinelearning,naturallanguageprocessing(NLP),and

predictiveanalytics.Itsmainproduct,anAIdataanalysisplatform,helpsorganizationshandlelargeamountsof

datainreal-time.Itcangenerateusefulinsightsandautomatedecision-making.Forexample,inhealthcare,

DeepSeek*salgorithmscandiagnosediseasesmoreaccurately.Infinance,itstoolscanpredictmarkettrendsand

reducerisks.

WhatmakesDeepSeekspecialisitsfocusonadaptabilityandethicalALThecompanytriestocreate

transparentAImodelsthatuserscantrust.Its"AdaptiveLearningFramework"allowssystemstochangeaccording

tonewdata,stayingusefulinchangingsituations.

Sinceitsstart,DeepSeekhasreceivedmuchpraise.IndustryleaderslikeitsabilitytoturnAItheoryinto

practicaluses.ACFOfromabigcompanysaiditcuttheiroperationalcostsby30%andimprovedforecasting.

Healthcareworkersalsolikehowitreducesdiagnosticerrors.ButDeepSeek'ssuccesshasalsostarteddiscussions

aboutAIethicsandworkforcechanges.

Inthefuture,DeepSeekplanstoworkwitheducationalinstitutionstotrainAItalentandstartcommunity

projectstosolveproblemslikeclimatechange.AsitsCEOsaid,theywantAItobringprogress,notjustprofit.

DeepSeekshowsthebrightfutureofAI,andit*sreadytochangethewayindustrieswork.

1.WhywasDeepSeekfounded?

A.Todevelopnewvideo-gametechnologies.

B.Tohelpindustriesmakeintelligentdecisions.

C.Tocreateasocialmediaplatform.

D.Tobuildself-drivingcars.

2.HowdoesDeepSeekhelpinthefinancefield?

A.Byprovidingfreefinancialadvicetoallusers.

B.Bypromotinghigh-riskinvestments.

C.Byincreasingthecomplexityoffinancialregulations.

D.Bypredictingmarkettrendsandreducingrisks.

3.WhatmakesDeepSeekspecial?

A.Itshigh-costproductsthatonlylargecompaniescanafford.

B.ItsfocusonadaptabilityandethicalAI.

C.Itslackofinterestinnewtechnologicaldevelopments.

D.Itspreferenceforworkingaloneratherthancollaborating.

4.WhatareDeepSeek*sfutureplans?

A.Toclosedownitscurrentbusinessandstartanewone.

B.Tofocusonlyonmakingmoreprofitwithoutanysocialcontributions.

C.ToworkwitheducationalfirmstotrainAItalentandstartcommunityprojects.

D.TostopusingAItechnologiesandswitchtotraditionalmethods.

Passage4

OnJanuary20,2025,aChinesetechfirmnamedDeepSeeklaunchedanewAI-DeepSeek-RI.This

intelligentprogramhastheabilitytosolvemathproblems,writecode,andanswerquestions,similartotop-tier

modelslikeOpenAPsGPT-40.However,itsdevelopmentcostismuchlower.Theteamonlyspent$5.58million

(1/70ofOpenAI'scost)andused2,048computerchipsovertwomonths,demonstratingthatgreatideascan

outperformlargeamountsofmoney.

DeepSeek-RIlearnsthroughrepeatedattemptsandself-improvement,muchlikestudentspracticingmath.

WhileotherAisrelyonnumerouspre-preparedexamples,RIcanenhanceitself.Forinstance,inaglobalmath

test,itachieveda79.8%accuracyrate,slightlysurpassingOpenAPsmodel.Moreover,abasicversionofRIcan

runonlow-costcomputers,benefitingschoolsandsmallenterprises.DeepSeek-RIisopenforfreeuseand

modification.Onitsfirstdayonline,morethan19,000programmersworldwideinitiatedprojectstostudyit.A

GermanscientiststatedthatexperimentswithRIcostonly$10,comparedto$300forotherAis.Evenatop

scientistfromMeta(awell-knownUStechcompany)said,"ThiskindoffreeAIcouldchangethe

world."

UStechcompaniesareconcerned.Metaplanstoinvestanadditional$65billioninAIresearch.Experts

believeChinaiscatchinguprapidly.AreportinNaturemagazinedescribedRIasa“bigstepforward",provingthat

smartideascandefeatexpensivetools.

AsaDeepSeekengineerputit,nWehopeAIwillbeafriend,notjustsomethingforthewealthytouse."With

RI,Chinashowstheworldthatgreatachievementscomefrominnovativeideasandhardwork,notjustmoney.

l.WhatisthemainadvantageofDeepSeek-RIcomparedwithothertop-tiermodels?

A.Ithasawiderrangeoffunctions.

B.Itcanrunonmoretypesofcomputers.

C.Ithasamuchlowerdevelopmentcost.

D.Itcansolvemorecomplexmathproblems.

2.HowdoesDeepSeek-RIlearnandimprove?

A.Byrelyingonalargenumberofpre-preparedexamples.

B.Throughrepeatedattemptsandself-enhancement.

C.BycomparingwithotherAImodelsconstantly.

D.Withthehelpofalargenumberofcomputerchips.

3.WhatdoweknowabouttheinfluenceofDeepSeek-RI?

A.Ithasattractedtheattentionofmanyprogrammersaroundtheworld.

B.IthasmadeUStechcompaniesgiveuptheirAIresearchplans.

C.IthasledtoareductionintheinvestmentofAIresearchintheUS.

D.IthasbeenrecognizedasthemostadvancedAIintheworld.

4.Whatisthebesttitleforthepassage?

A.TheDevelopmentofAIinChina

B.DeepSeek-RI:ANewMilestoneinAI

C.TheCompetitionbetweenChineseandUSAI

D.TheFutureofAI:DeepSeek-Ri'sVision

Passage5

（24-25高二上?貴州畢節(jié)?期末）ArtificialintelligenceischangingtheChinesefilmindustryalot.Ifs

makingbigchangesinhowmoviesaremade,fromcomingupwithideastowhathappensafterfilming.Areally

goodexampleofthischangeisashortmoviebyanexperienceddirectornamedYuBaimei.Thismovieshowsthe

famousChineselegendofZhongKuiinanew,modernway.ZhongKuiisaheroiccharacterfromoldChinese

stories.Usually,he'sshownasastrongwarriorinredclothes,fightingagainstbadspirits.

Yu,whois49yearsold,said,'TvealwayslikedZhongKui.He'sfunny,heroic,andweallknowhimwell.

ButitseemedlikehewasnevershownonTVorinmoviesinawaythatwasreallygoodenough.^^Withthehelpof

AI,Yumadeatwo-minuteshortfilm“NoDisturbancefromGhosts”.ThisfilmconnectstheoldstoryofZhongKui

toourmodernlife.

ThisspecialexperimentalfilmandothershortfilmsmadewithAIbyeightwell-knowndirectorswereshown

forthefirsttimeinDecember.Theywerepartofthe"KlingAIDirectorCo-CreationProject“attheChinaNational

FilmMuseum.Andtheseworkshavenowbecomepartofthemuseum'spermanent（永久的）collection.Thisisa

veryimportantandnewthingforChinesemovies.

Yusaid,"Soonerorlater,AIwillbepartofeverypartofourlivesandwillbelikeafriendtoeveryone.The

ninedirectorsinthisproject,nomatterhowgoodorbadtheirfinalworksare,havestartedsomethingveryvaluable.

Itdoesn'tmeanthatonlydirectorscanmakemoviesinthefuture,butthatanyonecancreate.^^

1.WhoisZhongKui?

A.Amodernhero.B.AChinesefilmmaker.

C.Aheroicfigureinoldstories.D.AcharacterinYuBaimei'slife.

2.Whydoestheauthormentionotherwell-knowndirectors5works?

A.Topromotethebestshortfilmsevermadesofar.

B.TostressAl'ssignificanceintheChinesecinema.

C.ToprovethevalueoffamousdirectorsinAIcreation.

D.Tosuggestthemuseum'slowstandardsforcollections.

3.WhatcanbeinferredaboutAIinthefuture?

A.Itwillreplacealldirectors.

B.Itwillonlybeusedinshortfilms.

C.Itwillmakefilm-makingmoreaccessible.

D.Itwillbetooexpensiveforcommonpeople.

4.Whatisthemainideaofthetext?

A.Heroicstoriesinthefilmindustry.B.Newprojectsoftrainingdirectors.

C.FamousdirectorsusingAIinfilms.D.AFsimpactonChinesefilm-making.

熱點(diǎn)話題02

閱讀七選五

Passage1

（24-25高三上?ill東濟(jì)寧?期末）Ifsnosecretthatartificialintelligenceischangingthewaywework.With

moreandmoresimplertasksbeingautomated,humansareleftwithmorehigh-levelcognitivework,suchasidea

generation,strategicplanning,andcreativewriting.]

Theproblemisthatwestillfollowamindsetthatresemblestheeraofassembly（組裝）lines,whichdoesn't

fosterorfacilitatethetypeofworkneededtoday.Leadersconstantlymeasuredproductivityintermsofhowmany

hoursanemployeespentatworkorthenumberofabstracttargetstheyreached.2Thistypeofoutput

doesn'tfitintheeraofALInstead,wehavetomakeapsychologicalswitch,structuringeachworkdayforquality

insteadofquantity.

Startbykickingtheoldmethodofcontinuousworkthroughouttheday.Researchshowsthatcontinuous

workimpairscreativeideagenerationandhighquality,complexcognitivework.3It'srecommended

thatusingworksessionsthatareabout90minuteslong,attackingthehardestpartofthetaskinthefirst20minutes

andmovingtotheslower,easierworkduringtheremainingtime.

4Therefore,managersshouldconsidertheprincipalareathatateamorindividualisworkingon

ratherthanimposingthesameworkscheduleacrossateamoranorganization.Datasuggeststhatit'sbettertowork

ontasksthatinvolvecreativity,imagination,andproblemsolvingduringthefirstfewhoursofthedayandagain

rightbeforeyougotosleep.5Inbetweenthefocusedworksessions,maketimeforanaporforeasy

workthatdoesn'trequiremuchcreativityorfocus.

A.Thafsbecausepeopletendtooverwork.

B.Thetimeofdaywillalsomakeadifference.

C.Asaresult,quantityofworkwasprioritized.

D.Toexcelinthenewera,we'llneedtotransformhowwework.

E.Witheverynewtechnologycomesfearandthennewopportunity.

F.Becauseofthat,ifsbettertoworkinbursts,givingthemindarest.

G.Themiddleofthedayisbetterforsomethingcallingforconcentration.

熱點(diǎn)話題03

語法填空

Passage1

DeepSeek,aremarkableChinese-madeAI,hasrecentlyburstintotheglobalspotlight.It1.(develop)

byagroupoftalentedChineseresearcherswiththeaimof2.(advance)artificialintelligencetechnology.

WhatmakesDeepSeekstandoutisitspowerfulcapabilities.Itcanhandleawiderangeoftasks,fromnatural

languageprocessing3.______imagerecognition.Innaturallanguageprocessing,itcangeneratehigh-qualitytext,

answercomplexquestions4.(accurate),andevenengageinintelligentconversations.Whenitcomesto

imagerecognition,itcanidentifyvariousobjectsandsceneswith5.(impress)precision.

ThesuccessofDeepSeekisnotaccidental.ItbenefitsfromthecontinuousinvestmentinAIresearchinChina

andtherichtalentpoolinthisfield.TheresearchersbehindDeepSeekhavemade6.(numerous)effortsto

improveitsalgorithmsandperformance.Theyhavealsoincorporatedthelatest7.(technology)

breakthroughsintothedevelopmentprocess.

Moreover,DeepSeekhasfound8.(apply)inmanyindustries.Inthemedicalfield,itcanassistdoctors

indiagnosingdiseasesbyanalyzingmedicalimagesandpatientdata.Intheeducationsector,itcanprovide

personalizedlearningresourcesforstudents.Inthebusinessworld,itcanhelpcompanieswithmarketanalysisand

customerservice.

AsDeepSeekcontinuestogainpopularity,itisplayinganincreasinglyimportantroleinpromoting

internationalcooperationintheAIfield.Itshowstheworldthe9.(remark)achievementsofChineseAI

technologyand10.(boost)theconfidenceoftheChinesescientificcommunity.

Passage2

閱讀下面短文，在空白處填入1個(gè)適當(dāng)?shù)膯卧~或括號(hào)內(nèi)單詞的正確形式。

WHATISDEEPSEEK?

OfficiallyknownasDeepSeekArtificialIntelligenceFundamentalTechnologyResearchCo,Ltd,thefirm1

(found)inJuly2023.Asan2(innovate)technologystartup,DeepSeekisdedicatedtodeveloping

cutting-edgelargelanguagemodels(LLMs)andrelatedtechnologies.

Sinceitsfirstmodel"DeepSeekLLM”releasedinJanuarylastyear,thecompany3(undergo)

multipleroundsofiteration.InDecember,thestartuplauncheditsopen-sourceLLMnV3,"4overtook

allofMeta'sopen-sourceLLMsandrivaledOpenAI'sclosed-sourceGPT4-O,accordingtoUSmediareports.

5traditionalapproacheslikeChain-of-Thought(CoT)andSupervisedFine-Tuning(SFT),

DeepSeekhasdistinguished6(it)intheAIindustrybyadoptingReinforcementLearning(RL)asacore

trainingmethod.

TheadoptionofRLhasallowedDeepSeektoenhanceitsmodels'reasoning,adaptabilityand7

(efficient),settingitapartasafrontrunnerinthefield.

When8(question)aboutthemeaningof"DeepSeek,"itslatestRIchatbotreplied,nThename

reflectsthecompany*smissiontodeeplyexplore9advancethefoundationaltechnologiesofAI,aiming

10(push)theboundariesofAIinnovationandapplication."

Passage3

閱讀下面短文，在空白處填入1個(gè)適當(dāng)?shù)膯卧~或括號(hào)內(nèi)單詞的正確形式。

OPEN-SOURCEMODEL

"ToseetheDeepSeeknewmodel,it'ssuperimpressiveintermsofbothhowtheyhavereally1

(effective)doneanopen-sourcemodelandthatitissuper-computeefficient,"saidMicrosoftCEOSatyaNadella.

Opensourceallowsresearchers,developersandusers2(access)themodel'sunderlyingcodeand

enablethemtouse,modifyorenhancethemodeltosuittheir3(need).DeepSeekhasgreatlybenefited

fromopen-sourceprinciplesand,4turn,demonstratesastrongcommitmenttosharingknowledgeand

contributingtothe5(collect)advancementoftechnology.

Meta'schiefAIscientistYannLeCunsaid:"Theycameupwithnewideasand6(built)themontop

ofotherpeople'swork.Becausetheirworkispublishedandopensource,everyonecanprofitfromit.”

EchoingLeCun,Sachs,theUSeconomist,said,nDeepSeek'sbusinessanddevelopmentmodelisopensource,

7isacompellingandsuccessfulmodelforscience,technologyandbusiness."

WhileOpenAIinitiallystartedasanopen-sourceorganization8latershiftedtoaclosed-source

model,DeepSeekhastakenadifferentpath.Highlightingtheimportanceof9(develop)collaboration

andinnovationthroughopen-sourceprinciples,LiangWenfeng,the10(founder)ofDeepSeek,said,

"Wewon*tchooseclosed-source."

Passage4

閱讀下面短文，在空白處填入1個(gè)適當(dāng)?shù)膯卧~或括號(hào)內(nèi)單詞的正確形式。

FinancialTimes(UK)《英國金融時(shí)報(bào)》

OnJanuary23rd,thenewspaperpointedoutthatDeepSeek's1.(succeed)isastrongtestamenttothe

growingindependentinnovationcapabilitiesofChina'sAL2.(reduce)itsrelianceonoverseaschips,

ChineseAIcompanieshavebeenvigorouslyinnovatinginkeyareassuchasalgorithms,significantly

3.(reduce)inferencecosts.Instead,importrestrictionshavespurred4.(bound)creativity.

BusinessInsider(US)《美國商業(yè)內(nèi)幕》

OnJanuary22nd,thereportstatedthatDeepSeek,5.ChineseAIstart-up,hasdemonstratedtheability

6.(catch)upwithtopAIlabsintheUS.ThenewlyreleasedmodelDeepSeek-RlbyDeepSeekisfully

7.(compare)toOpenAI'solmodel.ItspowerfulreasoningabilityhasgreatlyastonishedAmerican

8.(researcher),effectivelyprovingthattheUS*stechnologicalisnotsuperiorityunassailable.

Forbes(US)《美國福布斯》

OnJanuary22nd,thearticlesaidthattheUS'sexportcontrols9.advancedsemiconductors,originally

intendedtosuppressthedevelopmentofChina'sAI,have10.(expected)inspiredmoreinnovation.

DeepSeek'sRImodelappliesadvancedreinforcementlearningtechnologyandranksamongtheworld'smost

powerfulmodels,anditiscompletelyopen-sourced.ThisinnovativeapproachmakesDeepSeek*smodelsnotonly

powerfulbutalsocost-effective,promisingtopromotethepopularizationofadvancedAL

Passage5

閱讀下面短文，在空白處填入1個(gè)適當(dāng)?shù)膯卧~或括號(hào)內(nèi)單詞的正確形式。

Nature國際頂尖學(xué)術(shù)期刊《自然雜志》

StartingfromJanuary29th,itconsecutivelypublishedthreearticles1.(report)ontheimpactof

DeepSeekontheglobalacademiccommunity,including"ScientistsflocktoDeepSeek:Howthey'reusingtheAI

modelthat'stakingtheworld2.storm",nHowChinacreatedAImodelDeepSeektoshocktheworld",and

“China'scheap,openAImodelDeepSeekhasscientists3.(excite)".Thesearticlesshowedtheextensive

attentionandheateddiscussions4.(trigger)byDeepSeekwithintheglobalacademiccommunityfrom

differentperspectives.

Reuters(UK)路透社

OnJanuary31st,thereportsaidthattheDutchDataProtectionAuthority,aprivacy5.(regulate),stated

thatitwillinvestigateDeepSeek'sdatacollectionpractices6.urgedDutchuserstousethecompany's

software7.caution.

NHKWebsite日本廣播協(xié)會(huì)網(wǎng)站

OnFebruary1st,thereportsaidthatJapan*sDigitalMinister,AkiraTaira,statedthat8.thereare

concernsregardingthehandlingofpersonalinformation,etc.,hehopesthatpersonnelfrom

9.(variety)governmentdepartmentsusetheAImodeldevelopedbyDeepSeek10.(caution).

熱點(diǎn)話題04

書面表達(dá)

Passage1

假定你是李華，你的英國筆友Tom對人工智能很感興趣，他在郵件中詢問你關(guān)于最近在全球引發(fā)熱議

的DeepSeek相關(guān)情況。請你給他回一封郵件，內(nèi)容包括：

1.DeepSeek的基本信息(如它是什么，主要特點(diǎn)等)；

2.它在全球引發(fā)熱議的原因；

3.你對DeepSeek未來發(fā)展的看法。

注意：

1.詞數(shù)80詞左右；

2.可以適當(dāng)增加細(xì)節(jié)，以使行文連貫。

Passage2

學(xué)校英語俱樂部計(jì)劃舉辦一場關(guān)于人工智能發(fā)展的講座，此次講座將重點(diǎn)介紹在全球引發(fā)熱議

的DeepSeek。請你以俱樂部的名義寫一則通知，內(nèi)容包括：

1.講座的時(shí)間、地點(diǎn)；

2.講座的主要內(nèi)容（對DeepSeek的介紹，其影響等）；

3.歡迎大家參加并鼓勵(lì)提問交流。

注意：1.詞數(shù)80詞左右；

2.可以適當(dāng)增加細(xì)節(jié)，以使行文連貫。

<1

拓展培優(yōu)練

【同源語篇閱讀】

【材料來源：BBC]

DeepSeek:TheChineseAIappthathastheworldtalking

DeepSeek,aChineseartificialintelligence(AI)startup,madeheadlinesworldwideafterittoppedapp

downloadchartsandcausedUStechstockstosink.

InJanuary,itreleaseditslatestmodel,DeepSeekRI,whichitsaidrivalledtechnologydevelopedby

ChatGPT-makerOpenAIinitscapabilities,whilecostingfarlesstocreate.Itspopularityandpotentialrattled

investors,wipingbillionsofdollarsoffthemarketvalueofchipgiantNvidia-andcalledintoquestionwhether

Americanfirmswoulddominatetheboomingartificialintelligence(AI)market,asmanyassumedthey

would.PresidentDonaldTrumpdescribeditasa"wake-upcall"forUScompanies.

Whatisartificialintelligence?

TounderstandwhyDeepSeekhasmadesuchastir,ithelpstostartwithAIanditscapabilitytomakea

computerseemlikeaperson.Amachineusesthetechnologytolearnandsolveproblems,typicallybybeingtrained

onmassiveamountsofinformationandrecognisingpatterns.Theendresultissoftwarethatcanhaveconversations

likeapersonorpredictpeople'sshoppinghabits.

Inrecentyears,ithasbecomebestknownasthetechbehindchatbotssuchasChatGPT-andDeepSeek-also

knownasgenerativeAI.Theseprogramsagainlearnfromhugeswathesofdata,includingonlinetextandimages,to

beabletomakenewcontent.Butthesetoolscanalsocreatefalsehoodsandoftenrepeatthebiasescontainedwithin

theirtrainingdata.

MillionsofpeopleusetoolssuchasChatGPTtohelpthemwitheverydaytaskslikewritingemails,

summarisingtext,andansweringquestions-andothersevenusethemtohelpwithbasiccodingandstudying.

WhatisDeepSeek?

DeepSeekisthenameofafreeAl-poweredchatbot,whichlooks,feelsandworksverymuchlike

ChatGPT.Thatmeansit'susedformanyofthesametasks,thoughexactlyhowwellitworkscomparedtoitsrivals

isupfordebate.ltisreport