2025年金融服務(wù)領(lǐng)域生成式AI應(yīng)用實(shí)踐：機(jī)遇與風(fēng)險(xiǎn)管理研究報(bào)告（英文版）-UK Finance埃森哲

Incollaborationwith:

bUKFinance

GenerativeAIinAction:Opportunities&Risk

ManagementinFinancialServices

January2025

2UKFinanceLimited|GenerativeAIinAction:Opportunities&RiskManagementinFinancialServices

ThisreporthasbeenjointlyauthoredbyUKFinanceandAccenture,sponsoredbytheUKFinanceAIPolicyCommittee.

Primaryauthors(alphabeticallybysurname):

?SapanDogra(Accenture)

?MichaelErras(Accenture)

?CallumFarrell-Morris(Accenture)

?PeterHairs(Accenture)

?CarstenMaple(AlanTuringInstitute)

?WalterMcCahon(UKFinance)

?TomNiven(Accenture)

?BellaThornely(Accenture)

?LumaZitani(Accenture)

Contributorsandreviewers(alphabeticallybysurname):

?HeatherAdams(Accenture)

?ClaireAldworth(Accenture)

?MollyBloore(UKFinance)

?LiamCarlisle(Accenture)

?RayEitel-Porter(Accenture)

?SuhailKapoor(Accenture)

?AnnaKharchenkova(Accenture)

?EdKnight(Accenture)

?ChrisLane(Accenture)

?JoeLovatt(Accenture)

?AdamMarkson(Accenture)

?PhillipMind(UKFinance)

?JasonO’Brien(Accenture)

?LukaszSzpruch(AlanTuringInstitute)

?KuangyiWei(Accenture)

?MarkWelsh(Accenture)

?JenWilson(Accenture)

3UKFinanceLimited|GenerativeAIinAction:Opportunities&RiskManagementinFinancialServices

Contents

01

Reportoverview4

Foreword:UKFinance5

Foreword:Accenture6

ExecutiveSummary7

financialservices12

02

Today’sgenerativeAIlandscapeacross

Currentadoptionandnear-termtrends13

Sub-componentsofagenerativeAIsolution17

Prevalentusecasesinfinancialservices19

OverviewofgenerativeAIrelatedrisks21

Regulatorylandscape25

03

GenerativeAIcasestudies28

Casestudyone:Customercomplaintsagent30

Casestudytwo:KnowYourCustomer34

Casestudythree:Softwaredevelopmentlifecycle38

04

Keyrisksandmitigationapproaches42

Risktopicone:Reliabilityofoutputs43

Risktopictwo:Dataprivacyandsecurity47

Risktopicthree:Third-partyconsiderations50

05

Conclusionsandoutlook53

06

References55

4UKFinanceLimited|GenerativeAIinAction:Opportunities&RiskManagementinFinancialServices

overview

5UKFinanceLimited|GenerativeAIinAction:Opportunities&RiskManagementinFinancialServices

Foreword:UKFinance

ThearrivalofgenerativeAIhassparkedbothexcitement

andnervousnessamongthepublicandpolicymakers.While

enthusiasmispalpableatitsfluencyandabilitytoadapttoa

myriadoftasks,concernremainsaroundtherisksandpotentialforunforeseenissues.

Yetthefinancialsectorhasalonghistoryof

responsibleinnovation.Itsrecentadoptionofcloudtechnologystandsastestamenttohowfirmscansuccessfullyunderstandandmanageriskwhenembracingnewtechnologies.

AndtheexperiencessofarwithgenerativeAIindicatesthatthiscontinuestobethecase.

Thisreportshowsthatfirmsareinnovating

andexpandingtheirusecasesbeyond

chatbots,findingpracticalapplications

acrossdifferentfunctions.They’realsodoingsoprudently–expandingtheiruseofthe

technologyinstepwithimprovementstotheirtechnicalunderstandingandenhancementstoriskmanagementframeworks.

Ratherthanconsideringhypotheticaluse

casesthatmayariseatsomepointinthe

future,thisreportisfocusedonthenear-

term.Itisintendedtoilluminatehow

generativeAIisbeingusedinproduction

currentlyorwilllikelybeusedinthenear

future.Similarly,itexaminessomeofthe

mostchallengingfeaturesofgenerativeAI

andwhatmeasuresareavailable,oremerging,tohelpmanagetherisksposed.We’velookedcloselyatthreeusecasesandthreesetsof

risks,aimingtoexplorethesewellratherthantakingabroadbutshallowview.

JanaMackintosh

ManagingDirector,PaymentsandInnovation,UKFinance

Accenturehasbroughtawealthofexpertise

andknowledgetothisendeavour,whichwouldnothavebeenpossiblewithoutthem.Wearepleasedtohaveworkedtogether,alongsideourmembers,topublishthisreport.

Althoughthispaperisfocusedonthe

near-term,we’reexcitedtoseehowelse

generativeAIwillbeusedoverthemediumandlongterms.Inparticular,welookforwardtoseeinghowitcanbecombinedwithotherinnovations,suchaspredictiveAIorSmart

Data,toenablenewservices.

6UKFinanceLimited|GenerativeAIinAction:Opportunities&RiskManagementinFinancialServices

Foreword:Accenture

GenerativeAIhasreceivedahugeamountofattentionacross

boththebusinessandpublicdomain.SomeofthepredictionsforgenerativeAIarewell-founded,somearebalanced,andsomearerelativelyspeculativegivenhowquicklythetechnologyanditsuseinbusinessisevolving.

Regardlessoflonger-termpredictions,alotcanbelearntfromhowthefinancialservicesindustryhasapproachedthetopicoverthepasttwoyears.Firmsinthesectorhave

oftenbeenattheforefrontofexperimenting,innovatingandapplyingnewtechnologiestohowtheyruntheirbusinesses,whiledoingsoinaconsideredandcontrolledmanner.

GenerativeAIisnoexception.

WearedelightedtohaveworkedwithUK

Financeanditsmembersonthispaper.

Wehavedrawnonour‘ontheground’

projectexperienceofdeliveringAIsolutionsandstrategies,whileworkingcloselywith

academicresearchteams,UKFinance

membersandindustryresearchers.The

longer-termfutureofgenerativeAIhas

beenexploredinmanypapers,andwe’re

leftwithlittledoubtthatitwillhavea

significantimpactonfinancialservices.Thefocusinpreparingthispaperhasbeenon

thepracticalnear-tomid-termusesandthepragmaticwaysinwhichfirmsarenavigatingthepotentialrisks.

Thereporthighlightshow,inarelatively

shorttime,financialservicesfirmshave

demonstratedthetechnology’spotentialtoperformavarietyofcommontasksthatwereoncethedomainofhighlyskilledindividualsandteams.Wearealsoseeingexamples

ofgenerativeAIdeploymentsdrivingrealadoption,efficienciesandtangiblebenefits.

PeterHairs

ManagingDirector,

UKIAFinancialServices

Atthesametime,firmsareincreasingly

awareofthelimitations,risksand

uncertainties.Someoftheconsiderations

andconcernsassociatedwithgenerative

AIappearunprecedented,incontrastto

othertechnologiesthatweremorefamiliartoexistingteamsandbetteraddressedby

establishedriskmanagementapproaches.

Thishasleftbusinessesbothexcitedto

explorewhat’spossiblebutalsocautiousandmeasuredintheirapproach.

WehopethatthispaperhelpsfirmsontheiroverallAIjourneyandbringsabalanced

perspectiveastheynavigatetheopportunities,overcomepracticalhurdlesandrealisethe

potentialvalueofthisexcitingtechnology.

7UKFinanceLimited|GenerativeAIinAction:Opportunities&RiskManagementinFinancialServices

ExecutiveSummary

It’snowjustovertwoyearssincethelaunchofOpenAI’spublic

chatbotChatGPTon30November2022.ThiseventshowcasedtherapidprogressofgenerativeAItechnologiesinthepublicdomain.Sincethen,firmsacrossallindustrialsectorshavebeenadoptingandexperimentingwithgenerativeAI-basedtools,evaluatingthecapabilitiesofthetechnologyinthecontextoftheirownbusinessmodels,riskappetitesandorganisationalpractices.

Asahighlyregulatedsector,thefinancial

servicesindustryiscurrentlyworkingtoexploretheopportunitiesofferedbygenerativeAI,

whilecomplyingwithextensiveexistingandemergingregulationsandstandards.

Thisreportaimstoprovideafactualoverviewofthisemergingtechnologyinfinancial

servicesandisinformedbyUKFinancememberexperiences,expertdiscussionsandAccentureresearch.Itlaysoutsevennear-termusecasecategoriesprevalentintheindustrytodayandexplainshowtounderstandthedifferentlayersofa

generativeAIsystem.

Wehighlightthathumanexpertiseisatthe

coreofeffectivelyutilisingandcontrollingthetechnology,withsolutionstypicallyrelyingonhumanoversightfortraining,interpretation

andsensitivedecision-making.Alsoessential

isunderstandingthenatureofthemodels

underpinningthesesystemsandthe

dependencyonhigh-quality,structuredand

unstructureddatatoproduceaccurateinsights.

Identifyingnear-termopportunities

RealisingthevalueofgenerativeAIstartswithidentifyingthespecificbusinessfunctionsortasksthatthetechnologyiscurrentlysuitedto.Infinancialservices,generativeAIcanbeappliedacrossvariousfunctionsincluding

marketanalysis,financialcrimeandfraud

detection,customerserviceautomationandregulatorycompliance.

Manyrelatedpapersspeculateabouthow

generativeAIcouldrevolutionisetheindustry,automatecomplexactivitiesandtransform

customerexperience.Althoughmuchof

thisoptimismmaybewarrantedinthe

medium-term,theconditionstorealisethesepossibilitieshaveyettobeestablishedasweawaitfullyscalabletechnologicalfoundations,internaloperatingmodelsandgreaterclarityastopotentialregulatorychange.

Incontrast,thispaperfocusesonusesthatareeitheralreadydeployedtodayorareinadvancedstagesofdevelopment.Throughourassessmentoftheseusecases,the

prevalentuseofgenerativeAIisfocussedonsevenspecificareas:

?Customerengagementandpersonalisedmarketing

?Knowledgemanagementandinformationretrieval

?Softwaredevelopmentanddatamanagement

?Intelligentworkflowandemailprocessing

?Fraudandfinancialcrime

?Legal,contractualandcompliancetextanalysis

?Desktopandmeetingproductivity

8UKFinanceLimited|GenerativeAIinAction:Opportunities&RiskManagementinFinancialServices

AkeyobservationisthatmostlivegenerativeAIusecasesattheendof2024arefocused

onexploitingrelativelywellunderstood

capabilitiesofthetechnology,involve

activehumanoversightandarefocused

onrelativelylow-riskprocessesortasks.

Thisreportoutlinesthreeillustrativecase

studiesinmoredetailtohelpcharacterisethecurrentuseandprovidespecificinsights.

ReturnongenerativeAIinvestments

Illustratingthepathfromdiscussionandexperimentationtoimplementation,thelevelofinvestmentintogenerativeAI

fromfinancialservicesissteadilygrowing,

representing12percentoftechnology

investmentin2024,growingto16percent

in20251.GenerativeAIisbeingincreasinglyusedtooptimiserelativelydiverseprocesses,fromimprovingriskmanagementto

enhancingcustomerservice.UK-based

financialinstitutionsareinvestingin

generativeAItoautomatetheresource-heavytaskslistedabove.

Encouragingly,survey-basedindustry

researchsuggeststhatfirmsareincreasinglyseeingbenefitsfromtheirinvestment.

Satisfactionwithrealisedreturnon

investment(ROI)ishigh,rangingfrom75percentofexecutivesfromlargecorporationsto86percent2ofsmallandmedium-sized

businesses.Yettherearestillmanyfactorstoconsiderwheninvestingindeploying

generativeAI.Currently,thereislimited

abilitytoquantifytheROIandbaselinecostsrequiredtointroducethetechnology.3A

2024industrystudyhighlightedthatthe

mostcitedbarrierstofurtheradoptionwereimplementationcosts,qualityandaccuracyconcerns,datasecurityandprivacy,aswell

astrustandperceiveduseracceptance.

Despitetheseissues,it’sencouragingtoseefirmsareincreasinglybenefitingfromtheirinvestments.

Managinguncertainties

Perceivedregulatoryuncertaintyisanotherhurdletolarge-scaleadoption.Although

broadregulatoryapproachesarebecomingclearer,certainelementsanddetailsareyettobeconfirmed.Thelandscapecontinues

todeveloprapidly,andthelandmarkEUAI

Acthasgeneratedglobalattention.TheUK

bycontrasthassetoutitspro-innovation

approachinthe2023AIRegulationWhite

Paper.Thisoutlinesaprinciples-based

framework,insteadofawide-rangingAI

regulationintheEUstyle.ThefocusintheUKisoncontinuousupdatestosector-specific

regulationsasrequired,withtheFinancial

ConductAuthority(FCA)andBankofEngland(BoE)/PrudentialRegulationAuthority(PRA)responsesinApril2024confirmingthis.AnAIbillisplannedintheUK,butitsscopehasnotbeendetermined.

It’sthereforeessentialforbanks,insurersandassetmanagerstocontinueadaptingtheir

compliancestrategiestomeetestablishedregulationswhilealsodemonstratingtheeffectivenessofthesechanges.Those

whohavepreviouslyinvestedinstrong

complianceandriskframeworkswillbe

particularlywellpositionedforthisprocessandabletoinnovatesafely.

1Accenture-GeneratinggrowthhowgenerativeAIcanpowertheUK’sreinvention,2024

-analysisofanunpublishedfinancialservicessub-setofthedatarevealedthatthe2024

investmentis12%andtheexpected2025investmentwillbe16%ofoveralltechnologyspendongenerativeAI

2Google-TheROIofGenAI,Aglobalsurveyofenterpriseadoptionandvalue,n.d.

3Accenture-GeneratinggrowthhowgenerativeAIcanpowertheUK’sreinvention,2024-analysisofanunpublishedfinancialservicessub-setofthedata

9UKFinanceLimited|GenerativeAIinAction:Opportunities&RiskManagementinFinancialServices

GenerativeAI’srisks

GenerativeAIrequiresboththemanagementofrisksthatarewellunderstoodandothers

thatareheightenedinneworuniqueways.Tocategorisethese,wehaveutilisedtheNationalInstituteofStandardsandTechnology

(NIST)generativeAIrisktaxonomythathasemergedasamorematurestandardspecifictogenerativeAI.Wefocusedonthreerisk

topicsthatareamongthemostrelevanttothetypicalusecasesbeingexploredtodayandcriticaltofinancialservices:(1)accuracyofoutputs,(2)dataprivacy&securityaswellas(3)appropriateintegrationofthird-partysolutioncomponents.

Wediscusstheprevalenceoftheserisks,inadditiontomitigationapproachesusedinourcasestudiesandalaterin-depthsection

toinformthediscussionregardingsafeadoptionofgenerativeAIsolutions.

ThewideavailabilityofgenerativeAIalso

bringsrisksthatthetechnologycouldbe

misusedbybadactors,enhancingthreats

suchascyber-attacksorfraud.Whilethis

issueisimportant,thispaperfocusesinsteadontherisksassociatedwithgenerativeAIusebylegitimatecompanies.

IllustrativegenerativeAIcasestudies

Tohelpillustratereal-lifeapplicationswherefirmsareactivelydeployinggenerativeAI,

wehaveselectedthreecasestudiesthat

provideareasonablecrosssectionofcurrentgenerativeAIuse.Thesearecoveredinmoredetailinsectionfourofthisreport:

Casestudyone:Customercomplaints

Managingcustomercomplaintsisessentialformaintainingconsumertrustandregulatory

compliance.Theprocessinvolvesrecording,transcribing,investigatingandresolving

customercomplaints,ensuringissuesareaddressedfairlyandpromptly.Thislabour-intensiveandregulatedprocess,richindata,presentsasignificantcostandastrategicopportunityfordeployinggenerativeAI.

OnefirmhasfocusedondeployinggenerativeAItothisprocess.Followinganinitialpilotphase,generativeAIwasscaledtoproductiontosupport:

1.Producingcalltranscripts.

2.Summarisingkeyinvestigationfieldsfromvarioussources.

3.Analysingdocumentsprovidedbycustomers.

4.Draftingresponselettersincludingholdinglettersandfinalresponse.

5.Generatingpersonalisedfeedbackforagentsbasedoncomplaintresponse.

Benefitsincludedaproductivityincreaseof30-40percentandanimprovementinbothcustomerandemployeeexperience.

ThegenerativeAIsolutionwasnotgivendecision-makingpowers,whichrestedwithamemberofstaffwhoremainedaccountableforensuringfaircustomeroutcomes.Casemanagers’

knowledgewasactivelyutilisedtorefineandimprovemodelperformance.Inaddition,updatingoperatingproceduresandprivacydocumentationwereconsideredcriticalmeasuresfor

ensuringcustomersareinformedthatAIisbeingusedtosupportcasemanagerproductivity.

10UKFinanceLimited|GenerativeAIinAction:Opportunities&RiskManagementinFinancialServices

Casestudytwo:KnowYourCustomer

KnowYourCustomer(KYC)andCustomerDueDiligence(CDD)processesarefundamentalinfinancialservices.Theyaregovernedbystrictregulationsandcovertheentiretyofthe

customerlifecyclewithinthefirm,relyingonprocessingasignificantamountofstructuredandunstructuredpersonaldata.AgenerativeAIacceleratortoolwasdeployedtoprocessdocumentation,extractmandatedKYCinformationandpopulatethisintoanoutputformatthatwasmanagedbyexistingsystemsofrecord.Averyhighlevelofaccuracywasachieved.

Thistoolreducedprocessingtimesby

90percentforrelevantclients.

Tomanagerisks,thegenerativeAItoolwasalsocapableofrunningqualitychecksonthe

outputsbycomparingthemtosourcematerialandremediatingerrorsorblankfields.

Amanualqualitycheckwasthendonebyanoperator,whoassessedthetool’soutput

beforeconcludingtheprocess.Tomitigateprivacyanddatasecurityrisks,thesolutionwas

fullyhostedonaprivatecloudenvironment,withaprivateAPI(ApplicationProgramming

Interface)call-outtoaclosedLargeLanguageModel(LLM)inthefirm’sownprivateinstance.Additionally,accessrightsweretightlycontrolledanddocumentationencryptedatrestandintransit.DataminimisationwasappliedandthegenerativeAIenvironmentconfiguredforzeroretentionaftera30-dayperiod.

Casestudythree:Softwaredevelopmenttoolkit

Financialinstitutionsrelyheavilyontechnologytoruntheirbusinesses,drivestrategic

innovation,streamlineoperationsandenhancecustomerexperiences.Thesoftware

developmentlifecycle(SDLC)isapromisingareaforgenerativeAI-drivenoptimisationasaresult.TheSDLCisastructuredprocessusedbysoftwaredeveloperstodesign,develop,testanddeploysoftwareapplications.Theaimwastoaccelerateprogressthroughalarge-scaledatamigrationfromanon-premisesdatacentretothecloud.

ThefirmadoptedagenerativeAItoolkitfortherequirementsanalysisandtestingphases,

includinggeneratingsystemrequirementswithanLLMandfirm-specificinputs,aswellas

codeconversionandtestingaheadofhumanreview.Thedeploymentwasbasedonamulti-agentteamwherebythegenerativeAIsolutionadopteddifferentpersonas(e.g.designer,

developerandtester)tocritiqueandreacttoworkproducedbyotheragentsandraisethequalityofoutput.

ThistoolacceleratedtheseSDLCphasesbyover

50percentwithaccuracyover95percent.

Thismorecomplexarrangement,comparedtocasestudiesoneandtwo,reliedupon

additionalthirdpartiescontributingdifferentsolutioncomponents.Tomitigatetheassociatedthird-partyrisks,thecloudserviceproviderinvolvedguaranteedtheringfenceofclientdata

andcode.Specificprocessingcapacitywasalsoagreedtoenablelargermodelsandtrainingperformance,withServiceLevelAgreements(SLAs)underpinnedinthevendor’scontract.Previousmodelversionswerebacktestedtoavoidmodeldrift.Themodularapproach

enabledeffectiveorchestrationbyahumanintheloop(HITL),whoalsoassessedtheperformanceofeachgenerativeAIagent.

11UKFinanceLimited|GenerativeAIinAction:Opportunities&RiskManagementinFinancialServices

Conclusionandoutlook

Thereisameaningfulprogressionbeing

madefrommodestproofsofconcept(PoCs)withgenerativeAIsolutionstoreal-world

deploymentsthataredeliveringtangible

valueforbusinessprocesseswhilemanagingandmitigatingtheassociatedrisks.Firms

areinnovatingbutdoingsocarefully.

Thesedeploymentsillustrateagenerallyconservativeriskappetiteacrossthe

industry,ascanbeexpectedinahighlyregulatedenvironment.

Thesectorhasmanyyearsofexperience

insafelydeployinginnovativetechnology

andishometomaturegovernance,risk

andcompliance(GRC)capabilities.Financialservices,alongwiththeirtechnologyand

deliverypartners,havealsodemonstratedtheirabilitytoadaptthesecapabilitiestoemergingrisksandeventuallyintegrate

theirmanagementintobusinessasusual(BAU)processes.Thisismostrecently

demonstratedbytheexperiencewithrisksacrosscloudandcybersecurity.

Firmsshouldstartconsideringthese

capabilitiesasstrengths,equippingthesectorwithacompetitiveadvantage.Giventheir

trackrecordofsuccessfullyintegratingnewtechnology,financialservicesfirmsshouldfeelconfidentintheirabilitytoresponsiblyadoptgenerativeAIsolutionsandbegin

reapingthebenefits.

Thiswouldhelpfirmsacrossthesector

embracewider-rangingapplicationsthan

wehaveseensofarandenablescaled

adoptionacrossasecondwaveofusecases,inwhichgreatervaluecanbeunlocked

fromgenerativeAIsolutions.Forthisto

bepossible,firmswillneedtocontinue

toactivelyadoptgenerativeAIinaway

thatachievestheefficienciesandsavings

promisedbythetechnology,buildin-houseexpertiseandevolvetheirriskmanagementandgovernancelandscape.

Actionatindustrylevelcanenableand

accelerateresponsibleinnovationwith

generativeAI.Collaborationbetween

regulators,thefirmstheyoverseeandthoseofferingthesesolutionstothemarketcouldhelpclarifyareasofuncertainty.Similarly,

engagingcustomersearlytounderstandthelevelofacceptanceandaddresstheirconcernswillbekeytobuildingtrustandenablingadoption.

12UKFinanceLimited|GenerativeAIinAction:Opportunities&RiskManagementinFinancialServices

AIlandscapeacross

financialservices

13UKFinanceLimited|GenerativeAIinAction:Opportunities&RiskManagementinFinancialServices

Currentadoptionandnear-termtrends

Bylate2024,generativeAIhadevolvedfromanicheareaof

datascienceresearchintoafocalpointfortechnologyand

processinnovationformanyfirms.ThereleaseofChatGPT

in2022followedbyasuiteofenterprise-gradegenerativeAI

toolsin2023markedasignificantinflectionpoint,gathering

widespreadattentionacrossallindustries,includingfinancial

services.Furthermore,customerexpectationsandthecompetitivelandscapearelikelytoevolvebecauseoftheuseofAIin

everydaylife.

Manyorganisationsarenowgaininghands-onexperiencewithlowerriskandlower

complexityusecases,focusingonareaswheretheycandelivernetbusinessvaluewhilegettingtogripswiththetechnology.Thereisalsoanincreasedunderstandingofthespecificnatureandpracticalitiesofthetechnology,helpingfirmsfocustheirinvestmentsinAI.

IntheUK,investmentingenerativeAIhas

increasedsignificantlyoverthepastyear,

reflectingaglobaltrendtowardadopting

AI-driveninnovationsacrossindustries.TheUKgovernment,venturecapitalistsandtechcompanieshaveidentifiedthepotentialof

generativeAI,supportinginvestmentsin

research,startups,andinfrastructure.This

effortisbolsteredbythecountry'sstrong

academicecosystemanditsexpertisein

advancedAItechnologies.Thelaunchof

initiativesliketheNationalAIStrategyin2021highlightedtheUK’sstrategicgoaltobecomeagloballeaderinAI.

AI’spotentialisincreasinglybeingrealised

throughapplicationsinhealthcare,creative

industriesandfinance,withnotablegrowthininvestmentsfocusedonexpandingtheusesofgenerativeAI.

Manyfinancialinstitutionshavebeen

exploringandadoptinggenerativeAItosomeextentin2024andarelikelytoincrease

investmentsin2025.ArecentAccenture

surveyshowsamateriallevelofinvestmentaveraging12percentofthetechnology

budget,increasingtoapproximately16per

centoftotaltechnologyspendin2025.4,5

Whilesomefirmsfocusonspecificuses,38

percentofUKrespondentstoarecentsurveyhavedevelopedabroaderAI‘roadmap’

(ibid.).TheseincorporatemultipleprioritisedgenerativeAIandbroaderAI-basedsub-

initiatives,focusedonvalue,feasibility

andriskappetite.Whilethisconstitutesa

considerableproportionoffirms,lacking

thiscomprehensivelevelofplanningstill

constitutesaconstrainttoadoptionformany.

4Accenture-GeneratinggrowthhowgenerativeAIcanpowertheUK’sreinvention,2024

5Accenture-GeneratinggrowthhowgenerativeAIcanpowertheUK’sreinvention,2024

-Analysisofanunpublishedfinancialservicessub-setofthedatarevealedthatthe2024

investmentis12%andtheexpected2025investmentwillbe16%ofoveralltechnologyspendongenerativeAI,circa1%higherthancross-industryaverages

14UKFinanceLimited|GenerativeAIinAction:Opportunities&RiskManagementinFinancialServices

Figureone:ProportionoftechnologybudgetspentongenerativeAIin2024vs.2025,%

Source:Accenture

IntheUKfinancialservicesindustry,

generativeAIisbeingactivelyappliedin

multipleareasofday-to-da