AsiaPacificAIReadinessIndex20232.每日分享：6+份行研精選、3個行業(yè)主題2.每日分享：6+份行研精選、3個行業(yè)主題4.嚴(yán)禁廣告：僅限行業(yè)報告交流，禁止一切無關(guān)信息報告僅限社群個人學(xué)習(xí)，如需它用Contents1.1.ExecutiveSummary 1.1Background 2.MainFindings 2.1OverallAIReadiness 2.2GovernmentReadiness 2.3BusinessReadiness 3.Recommendations AppendixIMethodology AppendixIICorrelationAnalysisbetweenAIReadinessandEconomicPerformance....1921.ExecutiveSummary1.1BackgroundSince2019,whenthefirsteditionoftheAsia-PacificAIReadinessIndexwasreleased,artificialintelligence(AI)hasbecomeaneverydayrealityforconsumers,businesses,andgovernments.Thisacceleratedin2023withtheemergenceofgenerativeAI,whichhasbecomeoneofthemostimpactfultechnologylaunchesever,redefininghowweviewandrespondtotheimpactofAI.AccordingtoarecentMcKinseystudy,generativeAIalonecouldaddtheequivalentofUSD2.6trilliontoUSD4.4trillionineconomicbenefitsannually.1InAustralia,itisestimatedthatgenerativeAIcouldcontributeuptoAUD115billionannuallytotheeconomyby2030.2InJapan,itisestimatedgenerativeAIcanunlockUSD1.1trillioninproductivecapacity.3WithgenerativeAInowmorecapableandmorewidelyavailablethanbefore,consumersareusingitfortheirdailyinformationneeds,andcompaniesareincorporatingitintotheiroperations.Asaresult,two-thirds(67percent)ofITleadersareprioritisinggenerativeAIfortheirbusinesswithinthenext18months,4withone-third(33percent)claimingitasatoppriority.5Likewise,72percentofcompaniessaytheywillsignificantlyincreasetheirinvestmentsinAIoverthenextthreeyears,6and65percentofbusinessleadersindicatetheyareeitheracceleratingtheirexistingAIstrategiesorcreatinganAIstrategyforthefirsttime.7AtSalesforce,wearedeliveringapproximately200billionAI-poweredpredictionseveryday,upfrom6.5billionpredictionsinOctober2019.8SalesforceAICloudbringstogetherAI,data,analytics,andautomationtoprovidetrusted,open,real-timegenerativeAIthatisenterprise-ready.TheadoptionandutilisationofgenerativeAIrequiresalevelofAIreadiness,whichincludes:?Infrastructure:generativeAImodelscurrentlyrequiresignificantcomputationalresourcesorspecialisedhardware;?Data:generativeAImodelsrequirelargeamountsofhigh-qualitytrainingdatatolearnmeaningfulpatternsandgeneraterealisticcontent;?Skills:developinganddeployinggenerativeAImodelsrequireexpertiseinAIandmachinelearning;?Ethics:generativeAIcanraiseethicalconcerbiasedorharmfulcontent;and?Integration:AIreadinessalsoencompassestheabilitytointegrategenerativeAImodelsintoreal-worldapplicationsandworkflows.FordigitaleconomiestotakeadvantageofthefullpotentialofgenerativeAI,theyneedtohavestronginstitutional,infrastructural,organisational,andethicalfoundationsaroundAI.CommissionedbySalesforceandpreparedbyAccessPartnership,the2023Asia-PacificAIReadinessIndex(theIndex)aimstohelpAassessbusinesses’andgovernments’readinesstoadopt,deploy,andintegThis2023editioncovers12economiesintheregion:Australia,China(newaddition),India,Indonesia,Japan,Malaysia,NewZealand,thePhilippines,Singapore,SouthKorea(newaddition),Thailand,andVietnam.9Foreachcountry,theIndexmeasuresbusinesses’andgovernments’multi-facetedAIreadinessanditsimpactonsocio-economicopportunitiesthrough15statisticalindicators.10ThereportprovidesrecommendationsforbusinessesandgovernmentstohelpoperationaliseandmaximisetheuseofAI.Outofthe12economiescoveredinthe2023edition,five(Australia,Indonesia,NewZealand,Singapore,andThailand)AIreadinessimprovefromtheir2021score.ThisisadirectreflectionofthemanyAI-relatedinitiativesthoseeconomieshavelaunchedandimplementedbetween2021and2023.Singaporeholdsthenumberonespotforthethirdconsecutivetimesince2019,theresultofalargelyconducivepolicyandbusinessenvironment.Respectivelyranked2ndand3rd,JapanandChinahavedevelopedandimplementedseveralforward-lookinginitiativestoframeandenableAIadoptionbyallaspectsoftheeconomy.SouthKorea(4th)andAustralia(5th)—twoeconomiesthatarealsoprolificintheAIspace—closeoffthetopfive.2.GovernmentAIreadinessincreasedforGovernmentAIreadinesshasincreasedforallofthecoveredAPACeconomiesbetween2021and2023.ThailandandIndonesiaexperiencedthelargestincrease,atestamenttothemomentumcreatedbythepublicationoftheirlong-awaitednationalAIstrategiesin202211and2021,12respectively.EconomieshavegonebeyondrecognisingthevalueofAI;theyhavetakenactionandareeffectivelyreadytoharnessthetransformativepotentialofAIforthedigitalisationofgovernmentprocessesandservices.BusinessAIreadinesshasstagnatedformostAPACeco2021and2023.Thisassessmentislargelyduetotheinclusionofmicro,small,andmediumenterprises(MSMEs)inhaveatendencytodigitisemoreslowlyandatasmallerscalethanbiggercorporations.Vietnamexperiencedthelargestslowdown,andonly4.ThereisadividebetweeneconomieswithmatureandemergingOverall,the2023editionIndexshowsadividebetweenmatureandemergingdigitaleconomies:apartfromSingapore,allSoutheastAsiancountries—Indonesia,Malaysia,ThailanscoredbelowtheAPACaverage.HownationalAIpoliciesandstrategiesbetween2021and2022,suggestingthatitmayonlybeamatteroftimebeforewestartseeingtheon-the-groundimpactoftheseAIinitiativesonbusinessesandgovernmentorganisations.5.AIisadriverofeconomicdynamismandgrowthThisreport’scorrelationsshowthatanimportantfactorincountries’AIreadinessiswhethertheyhaveanopen,robust,dynamic,anddiversifiedeconomy.OverallAIreadinessstronglycorrelateswithGDPpercapita,theabilitytoleveragecomparativeandcompetitiveadvantages,thepropensitytoharnesstechnologicalinnovation,thedynamismofthelocalstart-upecosystem,andtheeasewithwhichbusinessescannavigategovernmentcatalystofchangeforAPACeconomieswhileeconomies’dynamismcanfurtherexpandthepossibilitiesofAIusecases.42.MainFindings2.1OverallAIReadinessTable1providesabreakdownofthescoresforthemaincomponentsoftheAIReadinessIndex.13SingaporeretainsthetoppositionacrossallformsofAIreadiness,followedbyJapaninoverallreadiness.Chinaranreadiness,andAustraliais2ndingovernmentreadiness.Sincethelastreportin2021,SingaporehaslaunchedmanydecisiveAI-relatedinitiatives,withanaimtoempowerpublic-andprivate-sectororganisationstoadoptAIinaresponsiblemanner:theNationalAIPrograminGovernment,14theNationalAIPrograminFinance,15theAIinHealthcareGuidelines,16andAIVerify.17Japan’sAIWhitePaper18anditsmultilateralpushfortheDataFreeFlowwithTrust(DFFT)framework19haveadvancedtheinternationalAIagenda,whileChina’s14thFive-YearPlanforNationalInformatisation20andWhitePaperonTrustworthyAI21haveenabledtheriseofmanyinnovativeAIcompanies.Likewise,AustraliahasreleasedtheNavigatingAIreport,22theAIEcosystemMomentumreport,23andtheResponsibleAINetwork24tooperationaliseAI.followedbyJapan(59.8),andChina(59.7).followedbyChina,(43.1),andSouthKorea(followedbyAustralia(77.7)andJapan(77.5).Table1:Overall,Busin(scoresoutof100)GovernmentAIReadinessGovernmentAIReadiness2023scoreSingapore(86.5)2Japan(59.8)China(43.1)Australia(77.7)3China(59.7)SouthKorea(42.7)Japan(77.5)4SouthKorea(59.2)India(42.2)China(76.3)5Australia(58.2)SouthKorea(75.7)6NewZealand(54.6)Australia(38.7)NewZealand(72.9)7India(49.8)NewZealand(36.2)Malaysia(64.4)8Malaysia(30.3)Thailand(59.9)9Thailand(43.6)Thailand(27.3)India(57.4)Indonesia(39.3)ThePhilippines(25.4)Indonesia(55.0)Vietnam(36.5)Vietnam(25.0)Vietnam(48.1)12ThePhilippines(35.7)Indonesia(23.5)ThePhilippines(46.0)BusinessAIReadiness2023scoreSingapore(53.6)OverallAIReadiness2023scoreSingapore(70.1)Ranking1Source:AccessPartnershipresearch5Japan2021#260.0#2#4#3Philippines2021#1038.52019#835.9#12#7MalaysiaIndia2021#651.1#1Indonesia2019#736.92021#165.62019#163.9#Japan2021#260.0#2#4#3Philippines2021#1038.52019#835.9#12#7MalaysiaIndia2021#651.1#1Indonesia2019#736.92021#165.62019#163.9#6Australia58.22021#357.82019#256.9LeaderAboveaverageSouthKoreaChinaThailand43.62021#940.32019#638.7Vietnam202143.3#11#92021#748.22019#546.4#10#8SingaporeNewZealand54.62021#554.0#5Figure1:OverallAIReadiness2019-202Lookingattrendsovertime,Figure1showFigure1:OverallAIReadiness2019-202?Singaporestandsoutbyconsistentlykeepingitstopspotsince2019,alongwithJapanretainingits2ndpositionsince2021.?TheadditionofChinaandSouthKoreatothe2023editionoftheIndexhaspushedAustraliato5thfromits3rdplacein2021.IndonesiaandThailandhaveseenthelargestriseintermsofoverallAIreadinessbetween2021and2023.TheintroductionofoverarchingnationalAIstrategiesbybotheconomies,alongwithasuiteofenablinggovernmentinitiatives.Overall,outofthe12economiescoveredinthe2023editionoftheIndex,five(Australia,Indonesia,NewZealand,Singapore,andseentheiroverallAIreadinessimprove..Belowaverage.Newin20216Japan2021#371.6#3#5#4Philippines46.02021#1140.42019#838.4#12#9MalaysiaIndia2019#452.32021#755.02019#550.2#1Indonesia2021#1043.3Japan2021#371.6#3#5#4Philippines46.02021#1140.42019#838.4#12#9MalaysiaIndia2019#452.32021#755.02019#550.2#1Indonesia2021#1043.3#6Australia2021#274.22019#269.1LeaderAboveaverageSouthKoreaChinaThailand2021#945.02019#642.8Vietnam2021#846.7#11#864.42021#657.1#10#7SingaporeNewZealand2021#566.3#2Figure2:GovernmentAIReadiness2019-2021-2023Figure2:GovernmentAIReadiness2019-2021-2023GovernmentAIreadinessassessesgovernments’abilitytoleverageandharnessdata-driveninnovationsbyexaminingtheextenttowhichtheybuildenablingpoliciesandconducivedegreetowhichAIisbeingusedbypublic-sectororganisationstoimproveefficienciesinthedetoarangeofdigitalexperiences.Figure2showsthatSingapore,Australia,andJapantakeupthetopthreespots,followedbyChina,SouthKorea,andIndia.Ranking10th,11th,and12threspectively,Indonesia,Vietnam,andthePhilippinescontinuetorankatthelowerendofthespectrum.Althoughgovernmentreadinesslevelshaveincreasedacrossalleconomies,thedifferencesbetweenhighandlowscorersremainsconsistentwiththe2021and2019editionsoftheIndex.VietnamhasshownthesmallestincreaseingovernmentAIreadiness,droppingthreespotsfrom8thspotin2021to11thin2023.WhiletheVietnamgovernmenthasrolledouttheNationalStrategyontheResearch,Development,andApplicationofAIUntiltheYear2030,25andhasbeenheavilyinvestinginAIandotherdigitaltechnologies,26progressisimpactedbyalackofAItalentandexpertise.27BelowaverageNewin202178Lookingcloseratthegovernmentreadinessindicators,Table2shows:?Acrossthespectrum,Singaporeleadsindigitaltransformationofthepublicsector,opengovernmentdata,humancapitalandresearch,ICTregulation,andgovernmentpromotionofinvestmentinemergingtechnologies.?Australiaisamongthetopscorers,reflectingitssustainedmomentumtobuildonitsrichopendatasources28andtoengagebeyondthepublicsectorthroughinitiativessuchastheAusGovernmentDataSummit.29Table2:GovernmentAIReadiness2023,detailedscores(scor?NewZealandisleadingindigitalgovernmentandopengovernmentdata.ThiscanbeattributedtotheOpenDataActionPlanrolledoutin2021,whichseekstoenableopendataandacceleratethereleaseandreuseofopengovernmentdata.30GovernmentAIReadiness2023CountryDigitalEvolutionIndexDigitalGovernmentScoreE-ParticipationOpenGovernmentDataIndex(OGDI)HumanCapitalandResearchH-indexforAIpublicationsICTRegulation(“Governance”pillar)GovernmentPromotionofInvestmentinEmergingTechnologies(“Government”pillar)Australia8.08.29.910.06.26.18.85.1China6.010.0IndiaIndonesia4.89.08.510.0NewZealand8.09.39.510.05.2Philippines0.86.03.9Singapore9.29.810.010.0SouthKorea4.84.4Thailand6.66.4Vietnam5.4Source:SeethefulllistofdatasourcesinAppendixI.Methodology.AllcalculationsbyAccessPartnership.APACdevelopmentsingenerativeAIThissectionprovidesanoverviewofthegenerativeAIlandscapeacrossthe12economiescoveredinthisreport.Australia:TheAustraliangovernmentandbusinesseshavebeentakingactivestepstoharnessthepotentialofgenerativeAI,whichisestimatedtoadduptoAUD115billionannuallytoAustralia’seconomyby2030.31InJuly2023,theDigitalTransformationAgency(DTA)andtheDepartmentofIndustry,ScienceandResources(DISR)publisof“interim”guidelinesontheuseofgenerativeAIbythepublicsector,forlowriskinstances.32Withintheprivatesector,AustralianemployeesareseeingariseintheuseofAIintheworkplace,withtwo-thirds(67percent)ofrespondentsusinggenerativeAItoolsatworkweekly.33China:TheCyberspaceAdministrationofChina(CAC)announcedcomprehensiveregulationsforgenerativeAIinJuly2023,emphasisingadherencetocoresocialistvaluesandacontrolledrolloutofChatGPT-likeservices.34Thenewrules,effectivefromAugust2023,coverallgenerativeAIcontentservicesofferedtotheChinesepublic.35India:AIisestimatedtounlockUSD621billionofproductivecapacityinIndia,whichisequivalenttoalmostafifthofthecountry’sGDPin2021.36India’sgenerativeAIfirmshaveattractedacumulativeinvestmentofmorethanUSD590million.37TheMinistryofElectronicsandIT(MeitY)hasannouncedplanstodraftanewroadmaptodevelopIndia’sAIecosystemfocusedonitsdomesticIndiaAIPlatformtopromotelocalstart-ups,research,andinnovation.38Indonesia:Indonesia’sgenerativeAImarketisforecastedtohaveacompoundannualgrowthrate(CAGR)ofover24.4percentfrom2023to2030.39SomebanksareexploringthepotentialofusinggenerativeAIcloudsolutionstoimprovethequalityoftheirchatbots,increaseemployeeproductivity,andautomatecontentcreation.40TheIndonesiangovernmenthasoutlineditsoverarchingAIpolicyframeworkintheStrategyforArtificialIntelligence2020-2045(StranasKA).41Japan:InApril2023,theLiberalDemocraticParty(LDP)releaseditsAIWhitePaper.42InMay2023,theJapanesegovernment’sAIStrategyCouncilreevaluatedandrestructuredtheissuesandstrategiesongenerativeAIin“TentativeSummaryofAIIssues”.43InJuly2023,thegovernmentannouncedaproposedAIauditandcertificationframeworkwhichoutlinesuserguidelinesforconfidentialityandmisinformationissuesandarisk-basedapproachtoensuregenerativeAIcompliancewithJapan’slawsandregulations.44Japan’sMinistryofEducationhaspublishedguidelinesallowingthelimiteduseofgenerativeAIinelementary,juniorhigh,andhighschools.45Malaysia:NearlyathirdofsurveyedcompanieshavealreadyintegratedgenerativeAI,andover40percentofICTprofessionalsreportthattheirorganisationsareactivelyexploringtheuseofgenerativeAItodrivebusinessgrowth.46TheMalaysiangovernmenthasoutlineditsnationalAIpolicyframeworkintheMalaysianNationalAIRoadmapfor2021-2025.47NewZealand:InMay2023,thePrivacyCommissionerofNewZealandreleasedguidanceonhisoffice’sexpectationsontheuseofgenerativeAIbybusinessesthataresubjecttothePrivacyAct2020.48ArevisedversionthatoutlinedpotentialprivacyrisksassociatedwithgenerativeAItoolswassubsequentlyissuedinJune2023.49ThePhilippines:ItisestimatedthatgenerativeAIcanunlockUSD79.3billionofproductivecapacity,whichisequivalenttoafifthofthecountry’sGDPin2022.50ThePhilippinesgovernmentlauncheditsNationalAIStrategyRoadmapin2021.51Singapore:InJune2023,theInfocommMediaDevelopmentAuthority(IMDA)publishedadiscussionpaper,incooperationwithAicadium,toshareSingapore’spracticalandaccretiveapproachtogenerativeAIgovernance.52InJune2023,DeputyPrimeMinisterHengSweeKeatsharedthatanewinitiative,ProjectMindForge,whichwouldhelpexaminetherisksandopportunitiesofgenerativeAIforthefinancialsectorbybringingkeystakeholdersfromthebankingsectortogetherwithleadingAIcompaniestopilotandprototypetoolsundertheauspicesofthegovernment.53SouthKorea:InMay2023,theSouthKoreangovernmentannouncedthatitwillsetnewstandardsandguidelinesforcopyrightsofAI-generatedcontentbySeptember2023tominimisedisputesandconfusionoverintellectualpropertyrights.54InJune2023,theNationalInstituteofTechnologyandStandards(NITES)establishedthefirstnationalstandard(KS)forAIethicsinKorea.55ThiswasaresponsetotheglobalproliferationofgenerativeAItechnologyandtheheighteneddiscussionssurroundingAIethics.56InJune2023,PresidentYoonSukYeoldeclaredtheParisInitiative,whichcontainsthebasicprinciplesforestablishingaglobaldigitalordertoestablishadigitalethicscodeintheeraofadvancingThailand:TheprivatesectorinThailandisexploringgenerativeAI’spotentialandapplicationsinmarketingandchatbots.58TheThaigovernmenthaspublisheditsNationalAIStrategyandActionPlan(2022-2027)inJuly2022.59Vietnam:TheprivatesectorhasshowninterestingenerativeAIapplications,butthenumberofgenerativeAI-focusedcompaniesremainsrelativelysmall.60In2021,theVietnamgovernmentlaunchedtheNationalStrategyontheResearch,Development,andApplicationofAIUntiltheYear2030.612.3BusinessAIReadinessBusinessAIreadinessassessesarefinancially,organisationally,andculturallycapableofadaptingtotherapidlychangingmarketdynamicsofanincreasinglydata-drivenglobaldigitaleconomy.TheBusinessAIreadinessscoreislowerthantheGoverinadditiontoenterprises.ItisestimatedthatSMEsalonecomprise96percentofallbusinessesinAsia.62Hence,eventhoughenterprisesarequicktoadoptAIintheirbusinessoperations,smallerplayerslikeMSMEslagbehindinadoptionduetolackoftechnologicalinfrastructure,budget,andashortageofAItalentandexpertise.63GenerativeAIislikelytoleveltheplayingfieldandencouragebroaderAIadoptionamongsmallerplayersbyofferingaffordableandcustomisedsolutions,automatingtasks,improvingdecision-making,enhancingcustomerexperience,andfacilitatingscalablegrowth.64?Singapore,China,andSouthKoreaholdthetopthreespots,followed10th,11th,and12th?Singapore,China,andSouthKoreaholdthetopthreespots,followed10th,11th,and12threspectively,thePhilippines,Vietnam,andIndonesiasitatthelowerendofthespectrum.?Singaporehasretainedthetoppositionintheregionsince2019,withandlogistics,andtourism—well-equippedtoadoptAI.65?InChina,AIadoptionhasmainlyfocusedonthefinancialservices,retail,andhigh-techsectors,whichtogetherrepresentoverathirdofthecountry’sAImarket.66?Vietnamhasshownthegreatestdecreaseinbdroppingfrom7thspotin2021to11thin2023.ThismaybeduetoashortageofAItalent.67?Ofthecountriescoveredintheinaugural2019editionoftheIndex,Indonesiahasexperiencedthelargestfall(droppingfrom8thspotin2019to12thin2023),followedbyMalaysia,Thailand,andthePhilippines(whichalldroppedbythreespotsbetween2019and2023).WhilethesecountrieshaveallrecentlylaunchednationalAIstrategiesandroadmaps,theyremainbelowtheAPACaverageforbusinessreadiness.Thishighlightsthepressingneedfornationalinitiativestoequiptheregion’s70millionSMEswiththetoolstocapitaliseonAI.68Japan2021#248.4#5#3#2Philippines2021#936.62019#733.5#10#4India42.22021#347.32019#245.2#1Indonesia2019#832.42021#149.7#7Australia2021#641.52019#344.8LeaderAboveaverageSouthKorea42.7ChinaThailand2021#1035.62019#634.6Vietnam2021#739.9#11#9Malaysia30.32021#839.42019#540.5#12#8SingaporeNewZealand36.22021#541.7#6BelowaverageNewin2021Lookingcloseratthebusinessreadiness?Singaporeisaheadofothercountriesintermsofbusinesses’adoptionofemergingtechnologiesandbusinesssophistication.Singaporealsoleadsintermsofventurecapitalavailabilityandvaluation,amajorfactorbehindthedynamism—andlongevity—ofAIstaknowledgeandtechnologyoutputs.69?SouthKorealeadsintermsofcreativeoutputs,whichincludeintangibleassets(suchasglobalvalueof5,000mostvaluablebrands),70creativegoodsandservices,andcreativegoodsexports(suchasmovi?IndialeadsintermsofnumberofAIstart-ups,scoringaheadofotherAPACeconomies.ThismaybelinkedtothefactthattotalinvestmentsinAIstart-upsinIndiastoodatUSD3.24billionin2022,withfundingdevotedtoAIstart-upstotallingUSD7.73billionbetween2013and2022.72GovernmentGovernmentAIReadiness2023CountryCompanies’AdoptionofEmergingTechnologiesBusinessSophisticationKnowledgeandTechnologyOutputsCreativeOutputsLabour-MarketReconfiguration(“Churn”)DuetoDigitalTransformationNumberofAIStart-UpsVentureCapitalAvailability&ValuationAustralia8.24.93.23.82.12.12.8China2.34.5India3.9Indonesia2.50.4NewZealand0.31.8Philippines2.60.1Singapore3.92.09.3SouthKorea2.3Thailand1.3Vietnam5.00.4Sources:SeethefulllistofdatasourcesinAppendixI.Methodology.AllcalculationsbyAccessPartnership.133.RecommendationsThe2023IndexfindingsshowthatdespiteAPACeconomies’varyingdegreesofAIreadinessandtheirdifferingapproaches,allofthemareprioritisingtheadvancementofAIandgenerativeAI.Comparedto2019and2021,AIisnowatthetopofnationalagendasforalleconomies,theimplementationofnationalAIstrategiesiswellunderway,andawarenessonthepotentialimpactofAIforeconomicgrowthanddevelopmentisatanall-timehigh.ImplementNationalAIstrategiesAtthetimeofwriting,mostAPACcountrieshaveaNationalAIstrategyinplace,and,insomecases,aNationalAIbodydevotedexclusivelytocoordinatingAIefforts.YetnotallcountrieshaveprogressedintermsofreadinesAIpolicyoragencyinplacedoesnotautomaticallyleadtoeffectiveAIadoptionandoperationalisation.Thisentailsgivinggovernmentagenciesthemandate,authority,legitimacy,budget,andresourcestoimplementthestrategy.Tobetrulyeffective,NationalAIstrategiesneedtobedistilledaswhole-of-governmentefforts,withactionplansandroadmapsthatareindustry-andsector-specific.AsdemonstratedbythethisisanapproachthatallowsnationalAIstrategiestobesustainedovertime.Adoptarisk-basedapproachtoAIregulation,andharmonisationandconsistencyaroundexistingrulesAtailoredrisk-basedAIregulationapproachiskey:Aone-size-fits-allapproachtoregulationmayhinderinnovation,disrupthealthycompetition,anddelaytheadoptionofthetechnologythatconsumersandbusinessesaroundtheworldarealreadyusingtoboostproductivity.73Risk-basedAIregulationfocusesmostonthehigh-riskapplications,especiallythosethatcouldcausesignificantharmorimpactsomeone’srightsandfreedoms.74Regulationshouldalsodifferentiatethecontext,control,andusesofthetechnologyandassignguardrailsaccordingly.Geinstance,shouldbeaccountableforhowthemodelsaretrainedandthedatatheyaretrainedon,whilethosedeployingthetechnologydecidehowthetoolisbeingusedandshouldhaverulesaccordingtothatinteraction.DataprotectionlawsthatprotectthefundamentalhumanrighttoprivacyareafoundationofresponsibleAIregulation.AIispoweredbydata.AdditionalrulesspecifictogenerativeAIshouldaddresstheuseandprivacyofpersonaldatafortrainingfuturemodels,safeguardingpersonaldatawithintheAIecosystem.75Finally,harmonisationandconsistencyshouldbeappliedwithexistingrules,someofwhichalreadyprovidesomeguardrailsaroundAI,suchasglobaldataprotectionlaws.Asregulatorsandotherstakeholdersdevelopnewguidance,theyshouldassessandclarifywhetherthereisanexistinglawaddressingtheseconcerns.76EnableresponsibleAIandethicsTheemergenceofgenerativeAIhasputthisissueattheveryforefront.Nearlysixin10(59percent)saidtheythinkgenerativeAIoutputsareinaccurate.77ResponsibleAIiscritical.Frompolicymakerstoentrepreneurs,thereisarealsenseofurgencytodevelopanddeployAIresponsibly.78ThedesignofresponsibleAIlieswiththeadoptionoffairnessprinciples,biassafeguards,andethicalguidelieverystageofAIdevelopment.Theseshouldbecomplementedbylaws,policies,andarisk-basedapproachgovdeploymentofAItoensureitsolvesproblems,improveslives,andleavesnoonebehind.TheUNESCORecommendationontheEthicsofArtificialIntelligence,theOECDPrinciplesonAI,andtheG20AIprinciplesareframeworksthatprovidecrucialguidanceonethicalandsocietalimplicationsofAI,caninformthedevelopmentofdynamicandsustainableAIecosystems.ItisimportanttonotethattherecentbreakthroughofgenerativeAIhasfurtherincreasedtheurgencytoenableresponsibleAI.OneareaforconsiderationforGovernmentistoappointaChiefEthicalHumaneUseOfficeroranAISafetyOfficertohelpwiththeimplementationoftheframeworkfortheethicalandhumaneuseoftechnologywithingovernmentandasitappliestodigitalgovernmentservices.Thecapa