mckinsey -在人工智能时代建立专业知识:谁培养下一代 Building expertise in the age of AI Who trains the next generation_第1页
mckinsey -在人工智能时代建立专业知识:谁培养下一代 Building expertise in the age of AI Who trains the next generation_第2页
mckinsey -在人工智能时代建立专业知识:谁培养下一代 Building expertise in the age of AI Who trains the next generation_第3页
mckinsey -在人工智能时代建立专业知识:谁培养下一代 Building expertise in the age of AI Who trains the next generation_第4页
mckinsey -在人工智能时代建立专业知识:谁培养下一代 Building expertise in the age of AI Who trains the next generation_第5页
已阅读5页,还剩17页未读 继续免费阅读

下载本文档

版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领

文档简介

Mckunsey

Quarterly

July2026

People&OrganizationalPerformancePractice

BuildingexpertiseintheageofAI:Whotrainsthenextgeneration?

AsAIreshapesentry-levelwork,organizationsshouldrethinkhowexpertiseisdevelopedbyintegratingknowledgemanagement,roledesign,learning,andcoachingintoasinglesystem.

byBryanHancock

withCharlotteSeiler

Fordecades,organizationshavereliedonearly-careertalenttodotheroutine,lower-riskworkthatsupportsthebusinessandtoserveasatraininggroundforfutureleaders.ThinkofPeggyOlson’strajectoryonMadMen,fromnoviceassistanttoDonDraper’sprotégétoconfidentcopychiefatanadvertisingagency—aclimbthatbeganbecauseroutineworkkeptherintheroom

wherejudgmentwaspracticed,andwhereherowncouldbenoticed.

Inreallife,advancesinautomationandAIarechangingthecompositionofentry-levelworkitself.Taskssuchasresearch,documentation,datacleanup,basiccoding,andpreliminary

analysisarebeingstreamlinedorabsorbedintoAIsystems.Thesearepreciselytheactivitiesthroughwhichyoungemployeeshavetraditionallybuiltinstincts,developedjudgment,and

earnedtherighttotakeonmore.

Atthesametime,fearsaregrowingabouttheimpactofAIonjobs.Inthe2025

Womeninthe

Workplacereport

byMcKinseyandLeanIn.Org,entry-levelworkers,particularlywomen,

reportedfeelingthemostworriedofallagegroupsabouthowAIusewillaffecttheirjobs.The

anxietyisbroadbased:Amonggraduatingseniors,pessimismaboutstartingacareerclimbedto62percent,from46percent,injusttwoyears,andthree-quartersofthepessimistspointedtofirmshiringfewerentry-levelworkersasthereason.

1

Theoutlookforentry-levelhiringisstillevolving,butearlyindicatorssuggestatightening

market.UnemploymentamongrecentUScollegegraduateshasincreasedsince2019,whilethenumberofentry-levelrolesisdeclining,particularlyinAI-exposedoccupations.Asofthefirst

quarterof2026,theunemploymentrateforrecentcollegegraduatesstoodatroughly5.7

percent,andaboutfourintenwereunderemployed—workinginjobsthatdonottypically

requireadegree.

2

ResearchfromtheStanfordDigitalEconomyLabfindsthatworkersages22to25“inthemostAI-exposedoccupationshaveexperienceda16percentrelativedeclinein

employmentevenaftercontrollingforfirm-levelshocks.”

3

HowmuchofthissofteningisattributabletoAIremainsgenuinelycontested:FederalReserve

BankofNewYorkeconomistsestimatethattheriseofremotework—whichmakesitharderto

trainnovicesatadistance—accountsformuchoftheincreaseinyoung-graduate

unemployment,andYale’sBudgetLabfindsnocleareconomy-wideAIfingerprintyet,evenasitflagsthegrowingdivergencebetweenyoungerandoldergraduatesasconsistentwithearly-

careereffects.

4

Whethertheapprenticeshipchannelisbeingerodedbyalgorithmsorbydistance,the

implicationforemployersisthesame:Theinformalmechanismsthatonceturnednovicesintoexpertscannolongerbetakenforgranted.Theseshiftspointtoanarrowingsetoftraditionalentrypointsjustasthenatureofearly-careerworkisbeingredefined.

Fororganizations,thisisamomentofbothuncertaintyandchoice.AsAIreshapeshowwork

getsdone,thequestionisnolongerhowmanyentry-levelrolestohire,but

whatthoserolesare

designedtodo

.Companiescanusethistransitiontorethinkentry-levelworkasafoundationforbuildingexpertiseinanAI-enabledenvironment—equippingearly-careeremployeestodesign,

1AIandtheworkforceahead:Whattheclassof2026tellsusaboutthefutureofthelabormarket,HandshakeNetworkTrends,2026.

2“Thelabormarketforrecentcollegegraduates,Q12026update,”FederalReserveBankofNewYork,May5,2026.

3ErikBrynjolfsson,BharatChandar,andRuyuChen,“Canariesinthecoalmine?Sixfactsabouttherecentemploymenteffectsofartificialintelligence,”StanfordDigitalEconomyLabworkingpaper,November13,2025.

4NataliaEmanuel,EmmaHarrington,andAmandaPallais,“Remoteworkleavesyoungerworkerssidelined,”FederalReserveBankofNewYorkLibertyStreetEconomics,June1,2026;“TrackingtheimpactofAIonthelabormarket,”TheBudgetLab,updated

June15,2026.

BuildingexpertiseintheageofAI:Whotrainsthenextgeneration?2

BuildingexpertiseintheageofAI:Whotrainsthenextgeneration?3

develop,andsteerAIsystems,notjustoperatewithinthem.

Thisarticleoutlinesafour-partapproachspanningknowledgemanagement,roledesign,

learningintheflowofwork,andmanagerialcoaching.Organizationsthatmovedeliberatelycanacceleratehowquicklyemployeesbuildjudgmentandtakeonhigher-valueproblem-solving,

strengtheningbothindividualexperienceandlong-termperformance.

Therewillalwaysbeapipeline

Entry-levelemployeesbringfreshperspectivesandcreativeproblem-solvingtoorganizations,fuelinggrowthandinnovation.Theyareapprenticestomore-seasonedmentors,building

relationshipsthatstrengthenthecultureforeveryone.Theybecomemiddlemanagersandseniorleadersdowntheroad,keepingthepipelinestrongintothefuture.

Thequestionofhowtogeneratethenextgroupoftalentisnotnew.Itis,however,being

exacerbatedasAIusecutsacrosssomanydifferentlevelsofknowledgeworksimultaneously.TwoseniorMicrosoftengineeringleaders,MarkRussinovichandScottHanselman,describethedynamicbluntly:Agenticcodingassistantsgiveseniorengineersan“AIboost,”multiplyingtheirthroughput,whileimposingan“AIdrag”onearly-careerdeveloperswholackthejudgmentto

steerandverifyAIoutput.Theresultingincentive—hireseniors,automatejuniors—quietlydismantlesthebottomofthetalentpyramidonwhicheveryseniorroledepends.Their

prescriptionisstrikingforitscandor:Keephiringearly-careeremployees,acceptthattheyinitiallyreducecapacity,andmaketheirgrowthanexplicitorganizationalgoal.5

McKinseysurveydatashowsthatgenAIuseisstartingtoaffecttheneedforentry-levelpositionsatsomeorganizations(Exhibit1).

Exhibit1

GenAluseisstartingtoaffectentry-levelrolesatmanyorganizations.

GenAl'simpactonneedforentry-levelroles,'%ofrespondents

Note:Figuresdonotsumto100%,becauseofrounding.

1Question:IsgenAIreducingtheneedforentry-levelrolesinyourorganization?source:Mckinsey'sNewEraofworksurvey,sept2025(n=28,000)

Mckinsey&company

5MarkRussinovichandScottHanselman,“RedefiningthesoftwareengineeringprofessionforAI,”CommunicationsoftheACMApril2026,Volume69,Number4.

BuildingexpertiseintheageofAI:Whotrainsthenextgeneration?4

Whilethatisatrendworthmonitoring,leaderswespeakwitharecommittedtohiringand

apprenticingthenextgenerationofexperts,eveniftheirnumbersarelowerthanbefore.And

someorganizationsareholdingtheirnumberssteady.BankofAmerica,forinstance,isbringinginnearly4,000summerinternsandfull-timecampusrecruitsin2026—matchinglastyear’s

hiringfromatalentpoolspanningmorethan500schools—whileexplicitlyredesigningthoserolesaroundAIfromdayone.Itschiefpeopleofficercallstheapproach“intentionalandlongterm.”

6

InaMarch2026surveyofroughly1,500executivesandseniortalentleaders,thoseexpectingAItoincreaseentry-levelhiringin2026outnumberedthoseexpectingittodecreasehiringbynearlythreetoone—thoughthesamesurveycarriesawarning,withathirdofemployers

reportingthatAIhasalreadyreducedthefoundational,skill-buildingtasksjuniorslearnfrom.

7

Organizationswanttohirepeoplewhohavethenascentskillstointerpretthein-depthdatageneratedbyAIandwhocanworkalongsideagents.Tocreateastrategythatguidestheseemployees,leaderscanfocusonfourareas.

Thefoundation:Knowledgemanagement

OrganizationsmustleveragetheirbestknowledgetodevelopAIworkflows.Iftheydon’tknowhowtomarshaltheircurrentexpertisetocreatehigh-qualitydata,itbecomesdifficulttofosterthenextgroupofexperts.Thegoalistomakeexpertiseusable,reliable,andembeddedinday-to-dayworkflows.

Codifiedexpertisecaptureshowtopperformersthink—theirframeworks,decisionrules,

assumptions,andpastjudgments—andstructuresthatknowledgesolargelanguagemodels(LLMs)canaccessit.Thisincludescleandata,curatedcasematerials,andstandardized

taxonomies.

Forexample,considerajuniorassociateatalawfirmworkingonanM&Acontract.Ratherthansimplysearchingacrossthousandsofpastdeals,anAI-enabledsystemcansurfacethemost

relevantprecedentsbasedondealcharacteristicsandhighlighttheclauses,trade-offs,andnegotiationpatternsusedbytoppartnersinsimilarsituations.Indoingso,itnotonly

acceleratesdraftingbutalsoexposesthereasoningbehindkeydecisions—helpingtheassociatebuildjudgmentwhiledoingthework.

Asimilarchallengearisesindeterminingwhichknowledgeshouldbeelevated.Athird-year

employeeatanengineeringfirmmaydocumentasingleclientengagementclearlyand

accurately.Butaseniorleader,drawingondecadesofexperienceacrossmanyprojects,may

produceanalysesthatreflectbroaderpatterns,trade-offs,andlong-termimplications.Both

inputsarevaluable,buttheyarenotequivalent.Forknowledgemanagementtobeeffectivein

anAI-enabledenvironment,organizationsneedwaystodistinguishbetweenisolatedexperienceandaccumulatedjudgment.

Thatdistinctionshouldbeexplicitinhowknowledgeiscuratedandsurfaced.Systemscan

6“BofAtowelcomenearly4,000summerinternsandcampusrecruits,”BankofAmericapressrelease,June3,2026;KatherineDoherty,“BankofAmericabucksAIjobfearswith2,000summerinterns,”Bloomberg,June3,2026.

7AndrewR.HansonandMollyCookEscobar,Entry-levelhiringintheAIera:Whatemployersarethinking(anddoing),StradaEducationFoundation,May19,2026.

BuildingexpertiseintheageofAI:Whotrainsthenextgeneration?5

weightinputsbasedonfactorssuchasdepthofexperience,repeatabilityofoutcomes,and

relevanceacrosscontexts.ThisensuresthatAItoolsandjunioremployeesareguidednotjustbywhatisdocumented,butbywhatreflectsthestrongestunderlyingexpertise.

Atitscore,modernknowledgemanagementindexesandprioritizesinformationinwaysthat

reflecthowexpertiseisbuilt,includingexpertvalidation,feedbackloops,andclearownership.Thesystemimprovesasitisused,capturingnewinsightsandrefiningwhat“good”lookslike.Thisenableslessexperiencedemployeestobuildjudgmentfasterandtakeonmorecomplexproblem-solvingearlierintheircareers.

Entry-levelroledesign:The‘answerkey’model

AsAIandagenticworkflowstakeonmoreexecutionofday-to-daybusiness,entry-levelrolescanberedesignedaroundworkingwith,supervising,andimprovingAI-drivenoutputs.This

createsanopportunitytoaugmentnear-termproductivitywhilepreservingthelong-termpipeline.

InanAI-enabledenvironment,thegoalisnolongerjusttaskproficiency,butjudgment:knowingwhentotrust,question,oroverridemachine-generatedoutputs.Whenknowledgemanagementandroledesignarestrong,early-careeremployeescanaccessinstitutionalexpertisefarearlier.Thelimitingfactorbecomestheirabilitytoapplythatknowledgeincontext.

AsMattBeanearguesinTheSkillCode,8earlyworkers’skillsdevelopthroughacombinationofchallenge,complexity,andconnectiontoexpertthinkingthroughapprenticeship—conditions

thatareoftenreducedwhentechnologyabsorbsroutinework.

Toavoidthisdisconnect,entry-levelrolescanincludeamoreintentionalformofapprenticeship,withAIintheloop.Teamscancreaterolesforlower-riskcontributionswhilebuildinginfrequentcoachingonhowtointerpretandrefineAIoutputs.Someorganizationsareintroducing

curriculum-basedroles,whereemployeesprogressthroughdefinedassignmentsthatpairhumanworkwithAIassistance.Othersarecreatingparallelworkflows,whereentry-level

employeescompleteworkindependentlyandthencompareitagainstAI-generatedresults,usingthegapasalearningtool.

Callthistheanswer-keymodel:Theemployeeattemptsfirst,theAIgradestheattempt,andtheemployeesitsdownwithamanager,whodiscussesthedifferences.Inthismodel,AIdoesnot

replaceapprenticeshipbutreshapesit,acceleratingfeedbackwhilepreservingtheconditionsrequiredtodevelopexpertise.

Atonerealestatefirm,anAIagentgeneratedhighlydetailedmarketassessmentsthatwere

morecomprehensivethanwhatanentry-levelemployeecouldproduceindependently.Ratherthanrelyingonthetoolalone,thecompanyaskedjunioremployeestobuildtheirown

assessments“byhand,”walkingneighborhoods,studyinggeographyandtrafficpatterns,and

forminganindependentview.TheythencomparedtheiranalysiswiththeAI-generatedoutput.Thecontrastservedasastructuredlearningtoolbyhighlightinggaps,surfacingmissedfactors,

8MattBeane,TheSkillCode:HowtoSaveHumanAbilityinanAgeofIntelligentMachines,HarperBusiness,2024.

BuildingexpertiseintheageofAI:Whotrainsthenextgeneration?6

andreinforcingstrongjudgment.Inthismodel,theagentacceleratesfeedbackandbroadensperspectivewhiletheemployeecontinuestodevelopthecontextualunderstandingand

instinctsthatunderpinexpertise.Usedthisway,AIsharpensentry-levellearning.

Evidencefromotherfieldssuggeststhecomparisonstepiswhatdoestheteaching.Inclinicalstudies,simplygivingphysiciansanLLMbarelyimprovedtheirlong-termdiagnostic

performance;aworkflowthatrequiredthemtocompareandreconciletheirownreasoningwiththatoftheAImodelliftedperformanceinfuturesituationstotheleveloftheAImodelalone.

9

Theinverseresultisjustasinstructive.WhenworkersusedgenAItoperformtechnicaltaskstheycouldnotdothemselves,thecapabilityvanishedthemomentAIaccesswasremoved.Nodurableskillhadformed.Passivereliancebuildsoutput;structuredcomparisonbuildsexperts.Onecaution:Workflowsthatdemandthiskindofcognitiveengagementarehardertouseandcanoverwhelmbeginners,sotheymustbedeliberatelydesignedanddosed,notboltedon.

10

Tomanagerisk,manyorganizationsareexpandingsandboxedenvironmentsforearly-careertalent.BankofAmericagivesinternsAItrainingspecifictotheirlineofbusinessfromdayoneandisusingsimulationtocompressthejudgmentbuildingthatjuniorbankersonce

accumulatedthroughroutinework.

“We’vegottogivepeopletheexperiencesinasimulatedwayquickly,”saysJoshBronstein,thebank’sheadofglobaltalent,“sothattheyhavethejudgmentthattheyotherwisewouldhave

gainedbydoingsomeofthosetasksintheirfirstyearortwo,thatnow,orsoon,AIorother

technologycando.”

11

Theseemployeescanpracticeininternalsimulationsorinexternal

contexts,suchasworkingwithnot-for-profitorganizations,beforeapplyingtheseskillsin

higher-impactsettings.ThisisincreasinglyimportantasjunioremployeesbegintoconfigureordirectAIagentsthatcanscaledecisionsquickly.

Theimplicationisthatentry-levelrolesarebecomingmoreaboutlearninghowworkgetsdoneinanAI-augmentedsystem—developingthejudgment,oversight,andadaptabilityrequiredtooperateeffectivelyasbothacontributorandsupervisorofdigitallabor.

Atthesametime,thestructureofentry-levelrolesthemselvesisshifting.Ratherthanconfiningnewhirestonarrow,task-basedspecialties,someorganizationsareredesigningrolesaround

end-to-endsystemthinkingandbroaderresponsibilities.

12

Insteadofhiringasalesanalyst,forinstance,acompanymayhireforabroaderrolethatencompassessales,marketing,and

commercialexpertise.Sincegeneralistshavethetoolstocustomizeforaspecificproblemorcontext,workcanbedesignedmoreforthem.

Don’twait:Learningintheflowofwork

Organizationscancreateagentsandlearningsystemsthatextendtheirdeepknowledgeand

trackskilldevelopmentinrealtime.Theanswer-keymodelevensuppliesthemetric:Thegap

betweenanemployee’sindependentattemptandtheAImodel’soutputisobservable,andagapthatnarrowsovertimeisdirectevidencethatjudgmentisforming—somethingapprenticeship,

9Gohetal.,“Largelanguagemodelinfluenceondiagnosticreasoning,”JAMANetworkOpen,October2024,Volume7,Number

10;Gohetal.,“GPT-4assistanceforimprovementofphysicianperformanceonpatientcaretasks:Arandomizedcontrolledtrial,”NatureMedicine,February2025,Volume31.

10JennaButleretal.,Microsoftnewfutureofworkreport2025,Microsoft,December2025.

11AliceTecotzky,“BankofAmericaacceptedlessthan1%ofinternshipapplicantsthisyear,butAIisn’tshrinkingtheclass,”BusinessInsider,June3,2026.

12AiliMcConnon,“ThebottomrungreturnsasAIreshapesentry-leveljobs,”IBM,March2,2026.

BuildingexpertiseintheageofAI:Whotrainsthenextgeneration?7

forallitshistory,hasneverbeenabletomeasure.Theideaof“learningintheflowofwork”

becomesmoreconcreteinthiscontextbecauseentry-levelemployeeslearnwhileproducingrealoutputs—withAIsystemsactingasembeddedguidesandreviewers.

AscompaniesredesignworkflowstoincludeAIagents,muchoftheroutineexecutionishandledbythesystem.Whatremainsforhumans,includingjunioremployees,ishighervalue:

interpretingoutputs,makingtrade-offs,andconnectinginsightsacrosscontexts.Thenatureofproductivitychangesasentry-levelworkerscontributeatahigherlevelearlierintheircareers.Forthistohappen,continuoussupportisbuiltintotheworkitself.AItoolscansuggestnext

steps,flagrisks,andprovideinstantfeedback,whilemanagersfocustheircoachingonjudgmentanddecision-makingratherthantaskmechanics.

Inpractice,someorganizationsfront-loadlearningthroughshortsimulationsorstructured

onboarding.Othersmovequicklyintorealwork,trustingthatAI-enabledfeedbackloopsandcoachingwillacceleratedevelopmentinrealtime.Inbothcases,themodelisthesame:

Employeesarenotwaitingtobecomeproductive;theyarelearningbydoing,withAIcompressingthetimeittakestobuildexperienceandearntrust.

Thelearningsciencebehindtheanswer-keymodelisbeginningtofirmup.Inoneexperiment,first-yearmedicalstudentswhoansweredAI-generatedcasequestionsandreceived

automated,personalizedfeedbackoverfivedaysoutperformedsecond-yearstudents—afull

yearaheadintraining—onthetargeteddiagnosesinavideo-basedexam,includingatatwo-

weekfollow-up.Thelearningpersisted,providingevidencethattheattempt-then-checkloop

canproducedurableskillratherthanborrowedcompetence.

13

AStanfordpilotpushesthe

designfurther:Achatbotplaysthepatient,andthestudentinterviewsthepatient,commitstoadiagnosis,anddefendsit,andonlythendoesthesystemcritiquethereasoning.

14

The

sequencingisthekeypoint:Thestudentgoesfirst,andAIgradestheattempt.

Hireforjudgment,coachforcontext

Organizationsarethinkingdifferentlyaboutwhotheyshouldhire.Insteadofbuildinglargecohortsofjuniorspecialists,theyareplacinggreateremphasisonadaptabilityovernarrowexpertise.Thequestionisshiftingfrom“Whatdidyoustudy?”to“Howdoyouthink?”

Inadditiontodigitalliteracy,employersareprioritizingattributessuchascreativity,problem-

solving,resilience,andabilitytoreason—qualitiesassociatedwithhighgeneralcognitiveability.

15

Thispatternshowsupinhiringdata:Whenemployersratetheskillsthatmattermostforentry-levelhires,theseskillsareatthetopofthelist(Exhibit2).

Thedurablebet,inotherwords,isjudgment,nottoolfluency.RolesthatdemandAIskillsare

alsonearlytwiceaslikelytodemandanalyticalthinking,resilience,oragilityalongsidethem.

16

OneheadofHRataFortune100financialfirmdescribedhiringnotforaspecificdegree,butfor“generalathletes”withstronglearninginstincts,relationalskills,andbaselinefamiliaritywithAI.

Thisshiftalsoopensthedoortoabroadersetofworkers,includingwhatthenot-for-profit

13YavuzSelimKıyaketal.,“AIteachessurgicaldiagnosticreasoningtomedicalstudents:Evidencefromanexperimentusingafullyautomated,low-costfeedbacksystem,”JournalofSurgicalEducation,October2025,Volume82,Number10.

14CarrieSpector,“StanfordeducationscholarusesAItohelpmedicalstudentshonediagnosticskills,”StanfordGraduateSchoolofEducation,September30,2024.

15AndrewR.HansonandMollyCookEscobar,Entry-levelhiringintheAIera:Whatemployersarethinking(anddoing),StradaEducationFoundation,May19,2026.

16ElinaMäkeläandFabianStephany,“Complementorsubstitute?HowAIincreasesthedemandforhumanskills,”arXiv,revisedJune8,2026,arXiv:2412.19754.

BuildingexpertiseintheageofAI:Whotrainsthenextgeneration?8

Exhibit2

organizationsareprioritizingskillsassociatedwithhighgeneralcognitiveabilityovertask-basedcapabilities.

TopfutureskillsaccordingtoHRprofessionals,2025-26,%ofrespondents(whomentionedtheskillamongtheirtop5)

skilltype:o——Traditionalo——Technological——Transformative——Digitalskillswiththe5largestincreasesinimportance

Note:year-over-yeardifferencesshouldbeinterpretedasdirectionalonly,ascountrycoveragechangedslightlycomparedwith2025,withtheadditionofBesource:MckinseyHRMonitorsurvey,Jan2026,n=1,303HRProfessionalsinBelgium,china,France,Germany,Italy,Netherlands,poland,spain,Uk,andUn=1,925HRProfessionalsinFrance,Germany,Italy,poland,spain,Uk,andUs

um,china,andtheNetherlands.

MckinseyHRMonitorsurvey,Dec2024,

Mckinsey&company

organizationOpportunity@Workcalls“STARs”(SkilledThroughAlternativeRoutes)—individualswithoutfour-yeardegreeswhohavebuiltcapabilitiesthroughexperience.Becauseknowledgeismoreaccessible,theseworkerscancontributemeaningfullyifgiventherightsupportand

developmentopportunities.

Thismarksadeparturefromthetraditionalhiringmodelthatrewardeddeepspecializationearlyon.AsAIandknowledgesystemsincreasinglyprovideon-demandexpertise,thevalueof

employeesliesinhoweffectivelytheyapply,connect,andextendthatknowledge.Inpractice,companiesareselectingmorewell-rounded,high-potentialindividuals—andrelyingon

apprenticeshipandreal-worldexperiencetobuilddomainexpertiseovertime.

Becausejunioremployeesarebeingpushedintohigher-valueactivitiesmuchearlier,coaching

BuildingexpertiseintheageofAI:Whotrainsthenextgeneration?9

forhumanskillsisevenmoreimportant.Anewhiremaybeinthepositionofengagingwith

seniorstakeholderssoonerthantheywouldhaveinthepast.Theymayhavetodeliverdifficult

messages,suchasrecommendingasignificantbudgetreduction,anddosowithjudgment,tact,andconfidence.Managerscanteachanewanalysthowtonavigateasensitiveconversation,

buildcredibility,orinfluenceanexperiencedoperator.

Withoutguidance,early-tenureemployeesmaydefaulttopresenting“whatthedatasays”inawaythatfeelsbluntordisconnectedfrombusinessreality.Managershavealwayscoached

youngeremployeestoaskthoughtfulquestionsandincorporateotherperspectives.Butthesetraitsareevenmoreimportanttoday,sincetheseemployeescaninterpretmorepotentand

comprehensivedatathaninthepast.FewseniorexecutiveshaveeverwantedtobetoldbyanewlymintedMBAthattheirbusinessplanisunrealistic.JunioremployeesmaybebackedbyAI,buttheystillneedtobecoachedtoreadtheroom.

Atthesametime,coachingcanfocusonhelpingnewhiresbuildcontextualunderstanding.As

analysisbecomesincreasinglyautomated,thedifferentiatorisnotproducingthenumberbut

explainingit.Newhiresneedtolearnhowtointerpretpatterns,recognizeindustrydynamics,

andarticulatethe“why”behindthedata.Thiskindofjudgmenttypicallycomesfromexperience,butinthisnewenvironment,coachescanacceleratethatlearningthroughreal-timefeedback,storytelling,andembedded“microlessons”thatexplainhowthebusinessactuallyworks.

Helpingananalystunderstand,forexample,thataspikeinsalesmaybedrivenbypromotionalcyclesratherthantruedemandisjustasimportantasteachingthemhowtopresentthe

numbers.

Somesuggesttakinginspirationfromthemedicalprofessiononformalizingcoachingintheflowofwork,givingittheinstitutionalweightthatinformalmentorshiphaslost.MollyKinder,an

expertonAI’simpactonwork,suggeststhatemployersborrowfromthemedicalresidency

model.IfAIabsorbstheroutinetasksthroughwhichjuniorsoncelearnedincidentally,

companiesshouldredesignentry-levelrolesasprotected,structuredperiodsofdeliberateskillbuilding,withprogressiontowardindependenceasanexplicitcommitment.

17

RussinovichandHanselmantakethatlogiconeleveldeeperintothedailyworkflow,proposinga“preceptor”model—aclinicalarrangementinwhichanewnurseorphysicianpracticesunderadesignatedseniorbeforeearningtherighttoworkindependently.

18

Intheirversion,asenior

engineerformallymentorsasmallcohortofjuniors,workingwithAItoolstogethersotheseniorcanobservewhatthejunioraccepts,rejects,andmisjudges—shiftingthementor’sjobfrom

answeringquestionstoteachingjudgment.

Ultimately,organizationsshouldtreatcoachingasacorecapabilityindevelopingearly-career

talent,notasanafterthought.Managersarenolongersimplysupervisingwork;theyare

responsibleforshapinghowemployeesthink,communicate,andengagewiththebusiness.Thismaybereinforcedthroughmorestructuredcoaching,betterknowledge-sharingsystems,and

redesignedentry-levelprogramsthatprovidebroaderexposuretodifferentfunctions.Asentry-levelrolesevolvefromexecutinganalysistoadvisingthebusiness,successwilldependonhow

17MollyKinder,“Tosaveentry-leveljobsfromAI,looktothemedicalresidencymodel,”BrookingsInstitution,January23,2026.

温馨提示

  • 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
  • 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
  • 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
  • 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
  • 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
  • 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
  • 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。

评论

0/150

提交评论