MicrosoftNew

FutureofWork

Report2025

Asummaryofrecentresearchfrom

Microsoftandaroundtheworldthat

canhelpuscreateanewandbetter

futureofworkwithAI.

MicrosoftNewFutureofWorkReportaka.ms/nfw

EditorsandAuthors

Editors:

JennaButler

(PrincipalAppliedResearchScientist),

SoniaJaffe

(PrincipalResearcher),

RebeccaJan?en

(SeniorApplied

Scientist),

NancyBaym

(PartnerResearchManager),

BrentHecht

(PartnerDirectorofAppliedResearch),

JakeHofman

(Senior

PrincipalResearcher),

SeanRintel

(PrincipalResearchSciencesManager),

BaharSarrafzadeh

(PrincipalAppliedResearchScientist),

AbigailSellen

(DistinguishedScientist),

MihaelaVorvoreanu

(PrincipalAppliedScientist),

JaimeTeevan

(ChiefScientistand

TechnicalFellow).

Authors:MohammedAlsobay,LizAnkrah,NancyBaym,StephanieBeers,MeganBenzing,MiaBruch,ZanaBu?inca,JennaButler,MarCarpanelli,AmeliaCole,ScottCounts,MadeleineDaepp,JustinEdwards,AlexFarach,DanGoldstein,MaryL.Gray,Brent

Hecht,JavierHernandez,JakeHofman,EricHorvitz,NicoleImmorlica,KoriInkpen,ShamsiIqbal,SoniaJaffe,ManasaJagadeesh,

RebeccaJan?en,SianLindley,BrendanLucier,NicMarquardt,MercyMuchai,AmbritaNand,AlexandraOlteanu,JackiO'Neill,MaxPeterschmidt,ChristianPoelitz,Rabeeza,NathalieRiche,SeanRintel,AdvaitSarkar,BaharSarrafzadeh,SunayanaSitaram,AmandaSnellinger,JinaSuh,JohnTang,LevTankelevitch,JaimeTeevan,KiranTomlinson,AnneTrapasso,AdamTroy,GauravVerma,MihaelaVorvoreanu,JackWilliams,JordanaYoung,BenZorn.

Referencingthisreport:

?Onsocialmedia,pleaseincludethereportURL(

https://aka.ms/nfw2025

).

?Inacademicpublications,pleaseciteas:Butler,J.,Jaffe,S.,Jan?en,R.,Baym,N.,Hecht,B.,Hofman,J.,Rintel,S.,Sarrafzadeh,B.,Sellen,A.,M.,Vorvoreanu,Teevan.,J.(Eds.).MicrosoftNewFutureofWorkReport2025.MicrosoftResearchTechReportMSR-TR-2025-58(

https://aka.ms/nfw2025)/2025.

Someoftheinformationinthisdocumentrelatestopre–releasedcontentwhichmaybesubsequentlymodified.Microsoftmakesnowarranties,expressorimplied,withrespecttotheinformationprovidedhere.Thisdocumentisprovided“as–is”.Informationandviewsexpressedinthisdocument,includingURLandotherInternetWebsitereferences,maychangewithoutnotice.Some

examplesdepictedhereinareprovidedforillustrationonlyandarefictitious.Norealassociationorconnectionisintendedorshouldbeinferred.ThisdocumentdoesnotprovideyouwithanylegalrightstoanyintellectualpropertyinanyMicrosoftproduct.

2

?2025Microsoft.Allrightsreserved.

MicrosoftNewFutureofWorkReportaka.ms/nfw

Welcometothe2025MicrosoftNewFutureofWorkReport!

Asyousitdowntoreadthe2025NewFutureofWorkreport,it’sworthpausingtoconsiderthethreadthattiesthepastfiveyearsofreportstogether.TheinauguralNewFutureofWorkreport,publishedin2021,focusedonnew

wayspeoplecouldworkwithoutrelyingoncolocationasakeyproductivitytool.Thesecond,in2022,centeredonthereintroductionofphysicalofficesandtheemergenceofhybridwork.In2023,weexploredhowlargelanguagemodelscouldreshapeeverydaywork,and,in2024,howthoseadvancesmovedfrompromisetoreal-worldimpact.

Eachyear,asI’vewrittenthisintroduction,I’vefoundmyselfsayingthatthepreviousyearmarkedaonce-in-a-

lifetimegenerationalshift.Butafterfiveyears,it’sclearthatthereportsaren’tcapturingaseriesofseparate

revolutions.Rather,theyarechaptersinasinglestoryofthedigitalevolutionofcollaboration,eachrepresentingaphasethatbuildson,andisenabledby,whatcamebefore.

Lastyear’sreporthighlightedresearchshowingthatAIdeliverssubstantialgainsinindividualproductivity.Thenextfrontier,coveredinthisyear’sreport,iscollectiveproductivity:howteams,organizations,andcommunitiescangetbettertogether.AIcanbridgegapsoftime,distance,andscale,butonlyifbuiltcorrectly.WemustdesignAIto

supportsharedgoals,groupcontext,andthenormsofcollaboration,andthisrequiresnotjustnewtoolsbutnewwaysofworking.

Microsoft’smissiontoempowereverypersonandeveryorganizationontheplanettoachievemoreremainsa

stablenorthstarastheterrainshifts.Ifthepastfiveyearstaughtusanything,it’sthatthefutureofworkisnot

somethingthathappenstous,it’ssomethingwecreatetogether,asaresearchcommunity,asanindustry,andasapublic.Asalways,weinviteyoutojointhateffort,approachingitwithcuriosity,intentionality,andguidedby

evidence,sothenextchapterofworkisbetterforeveryone.

–JaimeTeevan/ChiefScientistandTechnicalFellow

4

MicrosoftNewFutureofWorkReportaka.ms/nfw

ThisreportisaproductofMicrosoft’sNewFutureofWorkInitiative

Microsofthasshapedinformationworkforover50yearsandtheNewFutureofWork(NFW)initiativehashelpedit

navigatethepastfiveyearsofprofoundtransformation.

WhiletheinitiativewasbornoutoftheCOVID-19pandemicandsubsequentshiftstoremoteandhybridwork,since2023ithasfocusedontheincorporationandgrowingroleoftoolspoweredbylargeAImodels.

Throughoutthistransformativeperiod,theNFWinitiativehasbroughttogetherresearchersfromacrossMicrosoftto

conductprimaryresearchandsynthesizeexistingfindings

fromtheliterature.Thegoalisnotmerelytopredictwhat

changesarecoming,buttoactivelycreateanewfutureof

workthatisequitable,inclusive,meaningful,andproductive.

ThefiftheditionoftheannualNFWreportrepresents

anotheryearofresearchandanalysis,addingtothegrowingbodyofknowledgeonAIandwork.Theevidenceand

insightspresentedherearenotmonolithicandrepresent

contributingresearchers’perspectivesratherthanMicrosoft’scorporateviews.Additionalresearchpapers,practicalguides,andwhitepapersareavailableat:

aka.ms/nfw

.

aka.ms/nfw

5

MicrosoftNewFutureofWorkReportaka.ms/nfw

Reportoverview

Thisreportprovidesresearch-backedinsightsintohowAIis(orsometimes,shouldbe)shapingwork.Someofthequestionsitaddressesinclude:

?AdoptionandUsage:Whatchangesareoccurringinadoptionandusage?Whatarethedriversandchallenges?Whatarethegaps?

?ImpactonWorkandLaborMarkets:HowisAIimpactingworkandproductivity?Howarejobsevolving?Is

generativeAIaffectingemploymentandwages?Wheremightagentsreshapemarkets?Whatrolesdoautomationandaugmentationplay?

?Human-AICollaboration:HowarethewayspeopleinteractwithAIchanging?Howcanhuman-AIcollaborationbeimproved?HowdoesAIusedifferacrossmodalitiesandtimeframes?

?AIforTeamwork:HowcanAIsupportteamsaswellasindividuals?WhatrolecanAIplayinteamsettings?WhatisneededtoeffectivelyintegrateAIintogroupworkflows?

?Thinking,Learning,andPsychologicalInfluences:WhataretheeffectsofAIoncognitionandthinking?CanAIbedesignednotjusttocreateusefuloutput,buttomakethepeoplewhoworkwithitsmarter?HowcanAIserveasaneffectiveclassroomtool?Isitpossibletomeasurepsychologicalorwell-beingimpactsfromAI?

?SpecificRolesandIndustries:HowisAIchangingworkforsoftwareengineers,programmanagers,researchers,andotherprofessions?

?ExternalVoices:WhatdoleadingscholarsoutsideofMicrosoftthinkarethemostcriticaltopicsinthisspace?

Introductionaka.ms/nfw

Keycontext:AIcapabilitiescontinuetoadvance,especiallydueto

reinforcementlearning

?Long-horizontaskcompletioncapabilityismeasurablyaccelerating.METR’s50%task-completiontimehorizonshowsfrontieragents’reliabletask

lengthhasbeenrisingexponentiallywithan~7-monthdoublingtime,

turning“agentprogress”intoaconcretecapabilitytrend(Kwaetal.,2025).

?Verifiablereinforcementlearning(RL)post-training(rewardingcorrect,

checkableoutcomes)enabledstronggainsonhardmath/codingstyletasksevenwhenstartingfromabasemodelwithnolabeledreasoningtraces.

(DeepSeek-AIetal.,2025).

?Scalabletest-timecomputeframeworksgainedtraction,withopen-weightmodelsachievingIOI2025gold-levelperformance,showingrepeatable

TaskCompletionHorizonfromKwaetal.(2025).

“morecompute→higherscore”curvesincompetitiveprogramming.(Samadietal.,2025).

?Multi-turnRLfortool-using/searchagentsisnowbeatingprompt-onlybaselinesbylearningfromexperienceover

longeractionhorizons;onalegaldocumentsearchbenchmark,anRL-trained14Bagentreports85%vs78%againstfrontier-classmodels,withadditionalgainswhenallowedmoreturns.(Kalyan&Andrews,2025).

?Multi-modelrouting+aggregationisshiftingfromone-shot“pickamodel”toanRL-trainedsequentialpolicythatcanthink,callmultiplemodels,andintegrateresponseswhileexplicitlyoptimizingperformance–costtradeoffs—improvingresultsacrossmulti-hopQAstyleevaluations.(Zhangetal.,2025).

Kwa,T.etal.(2025).

MeasuringAIAbilitytoCompleteLongTasks

.METR.

DeepSeek-AIetal.(2025).

DeepSeek

-

R1:IncentivizingReasoningCapabilityinLLMsviaReinforcementLearning

.arXivWorkingPaper.

Samadi,M.etal.(2025).

ScalingTest

-

TimeComputetoAchieveIOIGoldMedalwithOpen

-

WeightModels

.arXivWorkingPaper.

Kalyan,V.andAndrews,M.(2025).

ReinforcementLearningforLong

-

HorizonMulti

-

TurnSearchAgents

.arXivWorkingPaper.

6

Zhang,H.etal.(2025).

Router

-

R1:TeachingLLMsMulti

-

RoundRoutingandAggregationviaReinforcementLearning

.arXivWorkingPaper.

AdoptionandUsageaka.ms/nfw

InvestmentinandadoptionofgenerativeAIcontinuestogrow

?EffectsofandgainsfromgenerativeAIwillbemediatedbyadoption(Jaffeetal.,2024).

?In2024,GenerativeAIhad$33.9billioninglobalprivateinvestment—an18.7%increasefrom2023.Publicinvestmenthasalsogrown(Maslejetal.,2025).

?AIuseatworkhasbeenincreasing,butheterogeneous:

?EnterpriseChatGPTmessagesincreased8xinthepastyear(Chatterjietal.,2025).

?AGermansurveyfound38%ofemployedrespondentsusedAIforwork(Giering&Kirchner,2025).

?AsurveyofenterpriseleadersfoundthatusageandconfidencewerehighestinIT&ProcurementandlowestinMarketing/SalesandOperations;the

Tech/Telecom,ProfessionalServices,andFinanceindustriesareleading(Korstetal.,2025).

UseofAI-basedsystemsatworktorecognizeandprocess(weightedpercentages)fromGiering&Kirchner(2025).

?Alarge2024surveyofAmericansfoundmenweremorelikelythanwomen(29.1%vs23.5%)tousegenerativeAIforwork(Bicketal.,2024).

?Ontheconsumerside,inJune2025,ChatGPThadover700millionweeklyactiveusersglobally(Chatterjietal.,2025).

?Thegendergap(basedonfirstnames)ofconsumerChatGPTusersisgone,adramaticchangesinceearly2023when>80%weremale(Chatterjietal.,2025).

MicrosoftStudy:Jaffeetal.(2024).

GenerativeAIinReal

-

WorldWorkplaces

.

Maslej,N.etal.(2025).

TheAIIndex2025AnnualReport

.AIIndexSteeringCommittee,InstituteforHuman-CenteredAI,StanfordUniversity.

Chatterji,R.etal.(2025).

ThestateofenterpriseAI

.OpenAI

Giering,O.andKirchner,S.(2025).

Artificialintelligenceandautonomyatwork:empiricalinsightsfromGermany

.JournalforLabourMarketResearch.

Korst,J.etal.(2025).

AccountableAcceleration:GenAIFast

-

TracksIntotheEnterprise.

WhartonHuman-AIResearchandGBKCollective.

7

Bick,A.etal.(2024).

TheRapidAdoptionofGenerativeAI

.NBERWorkingPaper.

AdoptionandUsageaka.ms/nfw

OrganizationalAIadoptiondependsonemployeesasmuchasleaders

?Acrossindustries,theintentiontouseAIisinfluencedbysocialnormslearnedfromleadersandpeers(Kellyetal.,2023).

?Workerscanbereluctanttoadopttop-downmandatedAIproductsthatprioritizeefficiencyabovequalityand

creativity,underminingthetraditionalviewofhumansasthecorevaluedriverofbusinesses.ThisreluctancelimitsthesuccessofAIpilotprograms(Youngetal.,2025;Sharma,2025;Murire,2024).

?LeaderscanfacilitateAIadoptionthroughclearcommunicationsupportingAIuse,demonstratingtheirownlearning,andsettingrealisticexpectationsaboutwhatAIcanaccomplish(Carteretal.,2024;Tursunbayeva&Chalutz-BenGal,2024).

?AIproductsthatintegratehumanthinking,creativity,andexpertisewhileamplifyingtheirvaluecanpromoteadoptionwithoutraisingconcernsaboutreplacement(Alietal.,2025;Youngetal.,2025;Sharma,2025).Forexample,anAIassistantcanactasathoughtpartner,helpingusersexploreideasandconnectconceptsacrosstheirknowledgebase.

?SomeofthebestwaysanorganizationmightfindtouseAI“comefromtheedge,notthecenter”(Winsor,2024).

OrganizationscanfacilitateAIadoptionbycreatingsystemsandincentivesforemployeestosharehowtheyuseAIwithoneanother(Tursunbayeva&Chalutz-BenGal,2024;Winsor,2024).

?EmployeesaremorelikelytoexperimentwithusingAIandtosharethoseinsightswithotherswhentheyfeelsafeandtrusttheirorganizations(Tursunbayeva&Chalutz-BenGal,2024;Bankinsetal.,2021).

?Manyemployees,particularlyGenX,willnotadopttoolsthatmakethemconformtoawayofworking-theywantproductsthatareflexibleenoughtofitpersonalwaysofworking(Rozsaetal.,2023;Doblinger,2023).

Kelly,S.etal.(2023).

Whatfactorscontributetotheacceptanceofartificialintelligence?Asystematicreview

.TelematicsandInformatics.

MicrosoftStudy:Young,J.etal.(2025).TheFutureofEnterprise.Internal.

Sharma,R.(2025).

TheImpactofAI

-

GeneratedContentonHumanCreativityandOriginalThought:APsychologicalStudy.

APA.

Murire,O.T.(2024).

ArtificialIntelligenceandItsRoleinShapingOrganizationalWorkPracticesandCulture.

MDPI.

Carter,J.etal.(2024)

ToSucceedwithAI,AdoptaBeginner’sMindset

.HarvardBusinessReview.

Tursunbayeva,A.andChalutz-BenGal,H.(2024).

Adoptionofartificialintelligence:ATOPframework

-

basedchecklistfordigitalleaders

.BusinessHorizons.

Ali,D.etal.(2025).

AIAdoptionAcrossMission

-

DrivenOrganizations.

arXivWorkingPaper.

Winsor,J.(2024).

HowtoBeSystematicAboutAdoptingAIatYourCompany

.HarvardBusinessReview.

Bankins,S.etal.(2021).

Amultilevelreviewofartificialintelligenceinorganizations

.JournalofOrganizationalBehavior.

Rozsa,Z.etal.(2023).J

obCraftingandSustainableWorkPerformance:ASystematicLiteratureReview

.Equilibrium.QuarterlyJournalofEconomicsandEconomicPolicy.

8

Doblinger,M.(2023).

AutonomyandEngagementinSelf

-

ManagingOrganizations:ExploringtheRelationswithJobCrafting,ErrorOrientationandPerson

–

EnvironmentFit

.Front.Psychol.

AdoptionandUsageaka.ms/nfw

CenteringworkervoiceinAIdesignboostsproductivity,satisfaction,andskill

growth—drivingbothbusinesssuccessandworkerflourishing

?Workerinvolvementintechnologydesignpromotessustainableproductivityandjobsatisfaction.Historicalandcontemporaryresearchconsistentlyshowsthatwhenworkers’expertiseandperspectivesinformthedesignanddeploymentofworkplace

technologies,organizationsachievemoresustainableimprovementsinproductivityandwell-being(Trist&Bamforth,1951;Roethlisberger&Dickson,1939;Hackman&Oldham,1976).

?EthnographicandHCIresearchdemonstratesthatworkersadapttechnologyincreativeways,andthatparticipatorydesign—whereworkersareco-designers—resultsintoolsthatbetterfitrealworkflowsandfosterhigheradoption(Suchman,1987;Orr,1996;Awumeyetal.,2024;Ehn,1993;Doellgastetal.,2025).

?CombiningtechnicalandsocialscienceresearchmethodscancreateAIsystemsthatimproveworkerskillsandsatisfaction—notjustaccuracy—byembeddinghuman-centricmetrics,workers’values,andskill-buildingintotheirdesign(Bucinca,2025).

?Data-drivenworkplacemonitoring(telemetry)hasmixedeffectsandshouldbegovernedwithworkerinputforbestoutcomes.Whilemonitoringandalgorithmicmanagementcanboostshort-termoutput,theyoftenincreasestressanderodetrustunlessworkershelpdefinewhatismeasuredandhowdataisused(Pentland,2012;Wellsetal.,2007;Ajunwa,2023).

Trist,E.andBamforth,K.(1951).

SomeSocialandPsychologicalConsequencesoftheLongwallMethodofCoal

-

Getting

.HumanRelations.

Roethlisberger,F.andDickson,W.(1939).

Managementandtheworker

.HarvardUniversityPress.

Hackman,J.andOldham,G.(1976).

Motivationthroughthedesignofwork:Testofatheory

.OrganizationalBehaviorandHumanPerformance.

Suchman,L.(1987).

PlansandSituatedActions:TheProblemofHuman

-

MachineCommunication

.CambridgeUniversityPress.

Orr,J.(1996).

TalkingAboutMachines:AnEthnographyofaModernJob

.CornellUniversityPress.

Awumey,E.etal.(2024).

ASystematicReviewofBiometricMonitoringintheWorkplace:AnalyzingSociotechnicalHarmsinDevelopment,DeploymentandU

se

.FAccT.

Ehn,P.(1993).

ScandinavianDesign:OnParticipationandSkill.

Doellgast,V.etal.(2025).

GlobalCaseStudiesofSocialDialogueonAIandalgorithmicmanagement

.InternationalLabourOrganization.

Bucinca,Z.(2025).

Worker

-

CentricAIforDecision

-

Support

.HarvardUniversity.

Pentland,A.(2012).

TheNewScienceofBuildingGreatTeams

.HarvardBusinessReview.

Wells,J.etal.(2007).

Theimpactoftheperceivedpurposeofelectronicperformancemonitoringonanarrayofattitudinalvariables

.HumanResourceDevelopmentQuarterly.

9

Ajunwa,I.(2023).

TheQuantifiedWorker

.CambridgeUniversityPress.

AdoptionandUsageaka.ms/nfw

CEOsexpectAItotransformtheirbusinesses,butleadingorganizationalAI

adoptioncanbechallenging

?A2025IBMsurveyof2000CEOsin33countriesand24industriesfoundthatmostCEOsexpectAItotransformtheir

businesses.OtherindustryresearchshowsleadersbelievehavingthemostadvancedgenerativeAIiscrucialtoremainingcompetitive(deBellefondsetal.,2024;IBMInstituteforBusinessValue,2025).

?However,organizationalleadershavedifficultiesdevelopingtop-downAIstrategiesformanyreasons,includingtherapid

diffusionofAItechnologies,thespeedwithwhichtheychange,theneedtocommunicateandreachalignmentaboutAI,theneedtoprioritizeAIagainstotherconcerns,andthechallengeofreimaginingworkflowsandprocesses(deBellefondsetal.,2024;Leonardi,2023).

?AcomparativecasestudyofAIadoptionintheDutchpublicsectorfoundadoptionbarriersincludedorganizationalinflexibilityandriskintolerance,aswellasstructuralseparationbetweenexplorationandexploitation.Forexample,adatascienceteam

mightbeputinchargeofexploringAIwithoutoperationalalignmentorfrontlineandleadershipsupporttoexploitandscaletheirideas(Selten&Klievink,2024).

?Chen&Tajdini(2024)surveyedmanagersinvolvedinMarketingandAIatUS-basedfirmsinvolvedinhigh-techconsumer

goods,industrialequipment,andfinancialservices.TheyfoundthatorganizationalintensityofAIadoptionwasdrivenbytopmanagementsupport,customerorientation,andemergingindustrysocialnormstowardadoptingAI(Chen&Tajdini,2024).

?Additionally,organizationsarebestabletoadoptAIwhentheyareinnovative,experimentalandlearning-oriented,supportive,andcollaborative(deBellefondsetal.,2024;Tursunbayeva&Chalutz-BenGal,2024,Sternfels&Atsmon,2025).

deBellefonds,N.etal.(2024).

Where’stheValueinAI?

BostonConsultingGroup.

IBMInstituteforBusinessValue(2025).

5mindshiftstosuperchargebusinessgrowth

.IBM.

Leonardi,P.(2023).

HelpingEmployeesSucceedwithGenerativeAI

.HarvardBusinessReview.

Selten,F.andKlievink,B.(2024).

OrganizingpublicsectorAIadoption:Navigatingbetweenseparationandintegration

.GovernmentInformationQuarterly.

Chen,J.andTajdini,S.(2024).

Amoderatedmodelofartificialintelligenceadoptioninfirmsanditseffectsontheirperformance

.InformationTechnologyManagement.

Tursunbayeva,A.andChalutz-BenGal,H.(2024).

Adoptionofartificialintelligence:ATOPframework

-

basedchecklistfordigitalleaders

.BusinessHorizons.

10

Sternfels,B.andAtsmon,Y.(2025).

Thelearningorganization:HowtoaccelerateAIadoption

.McKinsey.

AdoptionandUsageaka.ms/nfw

Socialnormsshapehowpeopleinterpretothers’(possible)AIuse,oftenwith

negativeconsequences

?ResearchhasshownthatsuspecteduseofAIcanresultinmoralevaluations.Peoplewho

useAIassistanceexpecttobe,andoftenare,evaluatedas“l(fā)azier”,lesscompetent,less

diligent,lesstrustworthy,andlessmoral(Reifetal.,2025;Schilke&Reimann,2025;Zhouetal.,2025).Theseeffectsarelessenedwhenevaluators,includingmanagers,haveexperiencewithandpositiveattitudestowardsAI(Reifetal.,2025;Schilke&Reimann,2025).

?BeingseenasusingAIcanharmrelationshipswithcolleagues,perhapsbecauseAI-usingemployeesmaybeseenas“slacking”ratherthanputtinginpersonaleffort(Zhouetal.,

2025).Paradoxically,theactofdisclosingtheuseofAIcanerodetrust(Schilke&Reimann,2025).

?Theextenttowhichthese“perceptualharms”(Kadomaetal.,2025)varyacrosssocialidentitiesisunclear:

DifferencesinevaluationsforAIhelpvs.Non-AIhelpandAIhelpvs.Control.Positivedvaluesindicatehigher

valuesintheAIHelpconditionwhilenegativedvaluesindicatelowervaluesintheAIHelpcondition.Errorbarsrepresent95%CI.FromReifetal.(2025)

?Reifetal.(2025)foundnogenderdifferencesinexperimentsacrossoccupationalcategories.

?Kadomaetal.(2025)foundthatmale(andEastAsian)freelancersweremorelikelytobesuspectedofusingAIandhenceevaluatednegatively.

?SoftwareengineerswhousedAIreceivedlowercompetencyratingsonidenticalwork,andthiseffectwasdoubledforfemales,withwomenreceivingacompetencyscore13%lessforidenticalcode(vs6%formen)(Gaietal.,2025).

Reif,J.etal.(2025).

EvidenceofasocialevaluationpenaltyforusingAI

.PNAS.

Schilke,O.andReimann,M.(2025).

Thetransparencydilemma:HowAIdisclosureerodestrust

.OrganizationalBehaviorandHumanDecisionProcesses.

Zhou,X.etal.(2024).

HowDoCoworkersInterpretEmployeeAIUsage:Coworkers'PerceivedMoralityandHelpingasResponsestoEmployeeAIUsage

.HumanResourceManagement.

Kadoma,K.etal.(2025).

GenerativeAIandPerceptualHarms:Who’sSuspectedofusingLLMs

?CHI.

11

Gai,P.etal.(2025).

CompetencePenaltyIsaBarriertotheAdoptionofNewTechnology

.SSRNWorkingPaper.

AdoptionandUsageaka.ms/nfw

TheResponsibleAIOrganizationalMaturityModelprovidesamapfor

organizationalchange

?AIadoptionandcapabilitybuildingrequireorganizationaltransformation(Kemp,2024).

Maturitymodelsareusefultoolsforguidingorganizationaltransformation.ExistingmaturitymodelsfocusonbuildingAIcapability(Alsheiabnietal.,2019;Vaishetal.,2021)andAI

adoption(Hansenetal.,2024).

?TheResponsibleAIOrganizationalMaturityModel(RAI-OMM)providesamapfor

organizationstoadvancetheirresponsibleAIstrategyandpractice(Hegeretal.,2025).TheRAI-OMMisforward-lookingandbestusedforplanning,notevaluation.

?Basedoninterviewsandco-designsessionswith90RAIexpertsandpractitioners,theRAI-

OMMidentifies24aspects(dimensions)thatneedtobeconsideredformatureRAIpracticeitiiin(eilI,-)fallintothree

anddescribes,foreachdimension,fivedistinctmaturitylevels.

?RAImaturityrequiresleadershipinvestment,alignedorganizationalpractices,andholisticchangemanagementstrategiesthataddressbothtechnologicalandhumandimensions(Duran,2025;Shekshnia,2025;Wangetal.,2025;Herrmann&

Pfeiffer,2023;Yunusa,2025).

?RAImaturitydimensionsareinterdependentandfallintothreecategories:OrganizationalFoundationsrequireleadershipcommitmentandinvestmentinorganization-wideinfrastructure;TeamApproachdimensionshighlightthenecessityof

cross-disciplinecollaboration;finally,thesemakepossiblematureRAIPractice,whichischaracterizedbydeepintegrationintoAIdevelopment&deploymentprocesses.

Kemp,A.(2024).

Competitiveadvantagethroughartificialintelligence:TowardatheoryofsituatedAI

.AcademyofManagementReview.

Alsheiabni,S.etal.(2019).

TowardsanArtificialIntelligencematuritymodel:Fromsciencefictiontobusinessfacts.

PACIS.

Vaish,R.etal.(2021).

AImaturityframeworkforenterpriseapplications

.IBMTechnicalReport.

Hansen,H.etal.(2024).

UnderstandingArtificialIntelligenceDiffusionthroughanAIcapabilitymaturitymodel.

InformationSystemFrontiers.

MicrosoftStudy:Heger,A.etal.(2025).

TowardsaResponsibleAIOrganizationalMaturityModel

.CSCW.

MicrosoftStudy:Duran,J.etal.(2025).

RAIadvocacy:CommunicativestrategiesforadvancingresponsibleAIinlargetechnologycompanies

.AIES.

Shekshnia,S.(