生成式AI时代的金融服务：促进负责任的使用

Contents Foreword2 Acknowledgements3 Executive Summary4 Foreword The report finds that the adoption of GenA.I. isprogressingsteadily across the financial servicesindustryin Hong Kong.However,there arechallenges hindering adoption, including concernsregardingmodel accuracy,data privacy andsecurity, as well as constraints related to resources GenerativeArtificialIntelligence(GenA.I.),arapidly evolving technology in its early stages ofdevelopment and adoption, represents the latestevolutionin Artificial Intelligence(AI).Globally Inthefinancialservicesindustry,GenA.I.applications have the potential to lead to morecurated customer experience, and more efficientwaysofprocessingandutilisingdigitalinformation.However,the adoption of GenA.I.could also give rise to new risks and challenges. Wehope the findings of this report canhelpinformbestpracticesforaddressingGenA.I.adoption challenges in the financialservices industry, and contribute to discussions onresponsible innovation and adoption, as well asindustry-widecapacity building.Looking ahead, Motivatedby recent developments,this reportprovidesacomprehensiveoverviewoftheevolution of GenA.I. and its broader implicationsfor both the financial services industry and financialregulators. The report draws on the findings fromasurvey and interviews commissioned by the Mr Enoch Fung Chief Executive Officer Executive Director Acknowledgements Wethank the Insurance Authority,MandatoryProvidentFund Schemes Authority,Securitiesand Futures Commission, as well as the BankingConductDepartment,theBankingPolicyDepartment, the Banking Supervision Departmentand the Digital Finance Division of the Hong KongMonetary Authority for their valuable commentsandsuggestions.We are also grateful for ourcollaboration with Ernst & Young Advisory ServicesLimited in designing and administering a survey, Executive Summary GenerativeArtificialIntelligence(GenA.I.),arapidly evolving technology in its early stages ofdevelopment and adoption, represents the latestevolutionin Artificial Intelligence(AI).GenA.I.models can learn from the patterns and structuresoftheir training data to generate outputs withsimilarcharacteristics,be it text,images,audio, There are a number of risk management challengeshindering adoption, including concerns regardingmodel accuracy, data privacy and security, as well asconstraints related to resources and talent.Whenadopting GenA.I.,financial institutionsconsideredmodel performance and accuracy(highlightedby 95%of surveyed firms),model Tostrengthenriskmanagement,financialinstitutions in Hong Kong have made solid firststepstowards responsible GenA.I.adoption anddevelopment,supported by updated regulatoryguidelines.There is a clear prioritisation oftransparency and accountability in GenA.I. tools,alongside a strong emphasis on data protectionand safeguarding customer information. Ongoing Theadoption of GenA.I.is progressing steadily across the financial services industry in Hong Kong.75%of the surveyed financial institutions havealready implemented at least one GenA.I. use case,or are currently piloting and designing use cases, GenA.I. adoption has been somewhat higher amongthe larger surveyed financial institutions.Amongsurveyed firms, 83% of large firms have rolled outat least one GenA.I. use case or are taking stepstowardsadoption,compared to 63%of smallfirms. Larger firms were typically more advanced in A ‘human-in-the-loop’ approach is also considerednecessaryto ensure proper risk managementand controls, especially at this stage of GenA.I.technology maturity. Consistent with this approach,about 80% of the surveyed financial institutionsidentifiedtechnical skills required in the use The primary implementations of GenA.I. in financialservices remain largely internal and non-customerfacing.75% of the surveyed financial institutionsviewed GenA.I. as a tool to enhance productivityand operational efficiency, followed by 53% whoviewed GenA.I. as empowerment for innovation Reskillingand upskilling appears to be a corepart of financial institutions’ talent strategy.Tohelpbridge skills gaps,the surveyed financial upskilling existing employees, hiring new talent,andestablishing external partnerships,especiallywithGenA.I. service providers. There is also anemphasis on continuous learning and innovation forfinancial institutions,especially those with asubstantial global footprint, as well as help prevent Based on these findings, the report outlines someconsiderations for facilitating responsible GenA.I.adoption by the financial services industry in HongKong.The Hong Kong government and financialauthoritieshave been active in undertaking amulti-pronged approach in supporting responsibleGenA.I.adoption and development,with the Theemergenceoflessresource-intensivemodelsand maturing technology,coupled withregulatoryengagement,are likely to contributetothe broadening of GenA.I.adoption overtime.The recent emergence of less resource-intensivemodels,suchasDeepSee