Generative AI's impact on bank relationship manager roles and customer lending experiences Vanessa Briglal Student Number: 2574300 Supervisor: Professor Thomas Dorson Anning A research proposal submitted to the Faculty of Commerce, Law and Management, University of the Witwatersrand, in partial fulfilment of the requirements for the degree of Master of Management in the field of Digital Business Johannesburg, 2025 ii KEYWORDS Generative AI, GenAI, Bank Lending, Relationship Manager (RM), Customer Experience iii TABLE OF CONTENTS List of Tables ................................................................................vi List of Figures ..............................................................................vii List of Acronyms ........................................................................viii CHAPTER 1. INTRODUCTION .................................................... 1 1.1 STATEMENT OF PURPOSE................................................................. 1 1.2 BACKGROUND OF THE STUDY ........................................................... 2 1.3 RESEARCH PROBLEM ...................................................................... 3 1.4 RESEARCH OBJECTIVES ................................................................... 5 1.5 RATIONALE .................................................................................... 6 1.6 DELIMITATIONS OF THE STUDY .......................................................... 7 1.7 DEFINITION OF TERMS ..................................................................... 8 1.8 ASSUMPTIONS ................................................................................ 9 1.9 CONCLUSION ............................................................................... 10 CHAPTER 2. LITERATURE REVIEW AND THEORETICAL FRAMEWORK 11 2.1 INTRODUCTION ............................................................................. 11 2.2 DEFINITION OF TOPIC OR BACKGROUND DISCUSSION .......................... 11 2.3 GENERATIVE AI ............................................................................ 13 2.3.1 DEFINITION............................................................................................... 13 2.4 TRANSFORMATIVE POTENTIAL OF GENAI .......................................... 16 2.4.1 AUTOMATION POTENTIAL – CREDIT ASSESSMENTS.......................................... 18 2.4.2 INNOVATION POTENTIAL - PRODUCT PERSONALISATION .................................... 20 2.5 RELATIONSHIP MANAGEMENT ......................................................... 21 2.5.1 EVOLVING ROLE IN THE ERA OF AI – IMPACT ON CUSTOMER EXPERIENCE ............ 22 2.6 GENAI IN RELATIONSHIP MANAGEMENT ............................................ 24 2.6.1 OPPORTUNITIES ........................................................................................ 25 2.6.2 CHALLENGES ............................................................................................ 27 2.7 ANALYTICAL FRAMEWORK .............................................................. 30 2.7.1 THEORETICAL FRAMEWORK ........................................................................ 30 2.7.2 CONCEPTUAL FRAMEWORK ......................................................................... 35 2.8 CONCLUSION ............................................................................... 49 iv CHAPTER 3. RESEARCH METHODOLOGY..............................50 3.1 RESEARCH PARADIGM ................................................................... 50 3.1.1 RESEARCH APPROACH ............................................................................... 52 3.2 RESEARCH DESIGN ....................................................................... 53 3.3 DATA COLLECTION METHODS .......................................................... 54 3.4 POPULATION AND SAMPLE .............................................................. 56 3.4.1 POPULATION............................................................................................. 56 3.4.2 SAMPLE AND SAMPLING METHOD .................................................................. 57 3.5 THE RESEARCH INSTRUMENT .......................................................... 60 3.6 PROCEDURE FOR DATA COLLECTION................................................ 61 3.7 DATA ANALYSIS STRATEGIES AND INTERPRETATION ........................... 62 3.8 POSSIBLE LIMITATIONS AND CHALLENGES OF THE STUDY .................... 65 3.9 QUALITY ASSURANCE .................................................................... 65 3.9.1 TRANSFERABILITY...................................................................................... 66 3.9.2 DEPENDABILITY ......................................................................................... 66 3.9.3 CREDIBILITY ............................................................................................. 67 3.9.4 CONFORMABILITY ...................................................................................... 68 3.10 ETHICAL CONSIDERATIONS ............................................................. 69 3.11 CONCLUSION OF RESEARCH METHODOLOGY ..................................... 71 CHAPTER 4. PRESENTATION AND DISCUSSION OF FINDINGS 72 4.1 INTRODUCTION ............................................................................. 72 4.1.1 RESEARCH OVERVIEW ................................................................................ 72 4.1.2 ANALYSIS FRAMEWORK .............................................................................. 73 4.1.3 THEME DEVELOPMENT PROCESS .................................................................. 73 4.1.4 CODE DEVELOPMENT AND ANALYSIS ............................................................. 74 4.1.5 PARTICIPANT DEMOGRAPHICS...................................................................... 79 4.2 THEME 1: IMPACT OF GENAI ON CREDIT LENDING .............................. 81 4.2.1 CHALLENGES ............................................................................................ 81 4.2.2 OPERATIONAL EFFICIENCIES ....................................................................... 85 4.2.3 PRODUCT PERSONALISATION ...................................................................... 91 4.2.4 ALIGNMENT TO THEORETICAL FRAMEWORK .................................................... 95 4.3 THEME 2: IMPACT OF GENAI ON RM ROLES ..................................... 98 4.3.1 TRANSFORMATIVE POTENTIAL ...................................................................... 98 4.3.2 IMPACT TO CUSTOMER EXPERIENCE............................................................ 104 4.3.3 ALIGNMENT TO THEORETICAL FRAMEWORK .................................................. 109 4.4 THEME 3: OPPORTUNITIES AND CHALLENGES OF INTEGRATING GENAI 111 4.4.1 OPPORTUNITIES ...................................................................................... 111 4.4.2 CHALLENGES .......................................................................................... 115 4.4.3 STRATEGIES FOR SUCCESSFUL INTEGRATION ............................................... 120 4.4.4 ALIGNMENT TO THEORETICAL FRAMEWORK .................................................. 131 4.5 CONCLUSION OF FINDINGS ............................................................133 v CHAPTER 5. CONCLUSION AND RECOMMENDATION .........135 5.1 INTRODUCTION ............................................................................135 5.2 RESEARCH OBJECTIVE 1 CONCLUSIONS ..........................................136 5.2.1 LITERATURE FINDINGS .............................................................................. 136 5.2.2 INTERVIEW FINDINGS ................................................................................ 137 5.2.3 GAPS IDENTIFIED IN THE LITERATURE .......................................................... 138 5.3 RESEARCH OBJECTIVE 2 CONCLUSIONS ..........................................139 5.3.1 LITERATURE FINDINGS .............................................................................. 139 5.3.2 INTERVIEW FINDINGS ................................................................................ 141 5.3.3 GAPS IN THE LITERATURE ......................................................................... 141 5.4 RESEARCH OBJECTIVE 3 CONCLUSIONS ..........................................143 5.4.1 LITERATURE FINDINGS .............................................................................. 143 5.4.2 INTERVIEW FINDINGS ................................................................................ 144 5.4.3 GAPS IN THE LITERATURE ......................................................................... 145 5.5 RECOMMENDATIONS ....................................................................147 5.5.1 RECOMMENDATIONS FOR BANKS ................................................................ 147 5.5.2 RECOMMENDATIONS FOR RMS .................................................................. 155 5.5.3 RECOMMENDATIONS FOR CUSTOMERS ........................................................ 159 5.5.4 RECOMMENDATIONS FOR POLICYMAKERS .................................................... 160 5.5.5 OVERALL STRATEGIC RECOMMENDATION ..................................................... 161 5.6 THEORETICAL CONTRIBUTIONS ......................................................161 5.6.1 HOW FINDINGS EXTEND UTAUT2............................................................... 162 5.6.2 NEW THEORETICAL INSIGHTS ..................................................................... 165 5.6.3 FRAMEWORK MODIFICATIONS .................................................................... 166 5.7 SUGGESTIONS FOR FURTHER RESEARCH.........................................169 5.8 CONCLUSION ..............................................................................172 REFERENCES ............................................................................174 APPENDIX ..................................................................................186 APPENDIX A1 - INSTRUMENT (INTERVIEW GUIDE FOR IT ARCHITECT) ............186 APPENDIX A2 - INSTRUMENT (INTERVIEW GUIDE FOR RMS) .........................187 vi List of Tables Table 1: Consistency table ............................................................................ 36 Table 2: Research proposition alignment to UTAUT2 framework .................... 48 Table 3: Participant demographics (Atlas.ti) ................................................... 59 Table 4: Structure of themes, categories and codes on Atlas.ti....................... 75 Table 5: Codes generated............................................................................. 76 Table 6: Document-category analysis ............................................................ 77 Table 7: Document-code analysis ................................................................. 78 vii List of Figures Figure 1: Causal Loop Diagram....................................................................... 1 Figure 2: Customer Journey Shift .................................................................... 4 Figure 3: Deciphering the Jargon around AI and GenAI ................................. 14 Figure 4: The impact of GenAI on different industries..................................... 17 Figure 5: Potential impact of GenAI across the lending journey ...................... 18 Figure 6: The unified theory of acceptance and use of technology2 (UTAUT2) 32 Figure 7: Causal Loop Diagram..................................................................... 35 Figure 8: Future Ready Business .................................................................. 39 Figure 9: Market for GenAI in customer relationship management .................. 42 Figure 10: C-Suite GenAI adoption concerns and challenges ......................... 46 Figure 11: Thematic Analysis Process ........................................................... 63 Figure 12: Word cloud view of coded data ..................................................... 76 Figure 13: Architecture vs RM feedback across categories ............................ 79 viii List of Acronyms ACRONYM DESCRIPTION 4IR The Fourth Industrial Revolution AI Artificial Intelligence DL Deep Learning IT Information Technology ML Machine Language NLP Natural Language Processing SME Small to Medium Enterprise TAM Technology Acceptance Model TOE The Technology Organisation Environment UTAUT2 Unified Theory of Acceptance and Use of Technology 2 1 CHAPTER 1. INTRODUCTION The chapter outlines the purpose of this study and the associated background that informed the research problem. It articulates the research objectives identified and the rationale for undertaking the research, followed by delimitations of the study. Key terminology used in this study is defined, followed by assumptions that have been made, ending with an outline of the chapter. 1.1 Statement of purpose This study investigates how South African banks can leverage generative AI (GenAI) technologies to enhance relationship managers' (RMs) customer service delivery in the lending sector. Employing qualitative methods, this study aims to understand how artificial intelligence (AI) can enhance customer interactions and drive organisational success. The diagram presented in Figure 1 illustrates a causal loop reinforcement model depicting the synergy between GenAI technology and its influence on RMs. Advancements in GenAI technology are posited to enhance the competencies of RMs, consequently improving the customer experience. This cycle suggests that customer feedback can inform iterative enhancements in GenAI systems. Figure 1: Causal Loop Diagram Source: Author’s own. 2 1.2 Background of the study The banking sector in South Africa is undergoing rapid digital transformation, driven by changing consumer expectations and technological advancements (Bansal & Chavva, 2024). With the evolving needs and expectations of tech savvy customers, banks are challenged to innovate and adapt to improve their offerings and stay ahead of the competition. While traditional AI has been transforming banking operations, the emergence of GenAI presents a distinct paradigm shift in how banks can optimise customer service processes and elevate the customer experience. Domokos and Sajtos (2024) and Donepudi (2017) describe AI as machines that are programmed to mimic human intelligence by analysing data to recognise patterns and make decisions. The authors suggest that GenAI is a subset of AI due to its ability to generate fresh content based on the learned data. In essence, while AI is limited to analysing existing data, GenAI can generate new content from patterns it has learned using AI. By leveraging advanced machine learning (ML) algorithms and natural language processing (NLP) capabilities, GenAI systems can analyse vast amounts of data, extract insights, and generate contextually relevant responses in real-time (ABSA, 2024). Research conducted by Accenture (Abbott, 2024), analysing over 19,000 banking tasks across 900 job families, found that 73% of banking activities could be enhanced through GenAI - 39% through direct automation and 34% through augmentation. These findings resonate globally, with South Africa included. Mossavar-Rahmani and Zohuri (2023) and Kalia (2023), on the other hand, argue the ethical dimensions of integrating AI into financial services, highlighting the critical need for transparency and accountability in AI systems, particularly in the context of addressing biases and inequalities. This focus on ethical dimensions sparks debate about the potential implications of tools like GenAI on the roles and responsibilities of individuals in financial institutions, notably RMs. 3 RMs plays a pivotal role in building strong relationships with customers, ensuring their satisfaction and loyalty throughout their interactions with the bank (Hamzah et al., 2016). They act as the main point of contact between clients and the institution; handling inquiries, resolving issues, and providing personalised financial advice (Zegullaj et al., 2023). By leveraging AI technology, these managers can deliver tailored solutions, anticipate customer needs, and enhance overall satisfaction and loyalty. Miah (2024) argues, however, that while AI tools can improve the efficiency and accuracy of services provided by RMs, they also bring challenges such as the requirement for new skills, potential job displacement, and ethical considerations in decision-making. Due to limited research on the tangible impact of GenAI in helping RMs provide improved customer service, particularly in bank lending, this study aims to fill this gap and enrich the current body of knowledge on AI technology within the banking industry. The research offers practical guidance for RMs on how to effectively leverage GenAI to elevate the overall lending customer experience. 1.3 Research problem In financial lending, delivering an exceptional customer experience hinges on RMs’ ability to accurately assess creditworthiness, offer personalised recommendations and streamline the lending application process (Zegullaj et al., 2023). These RMs, however, often encounter obstacles due to manual processes and outdated systems, leading to extended processing times and decreased customer satisfaction. Krause (2024) posited that the integration of Generative AI with other cutting- edge technologies is poised to reshape the financial landscape. Accenture (Abbott, 2024) corroborated, stating that GenAI technologies like ML and NLP could potentially revolutionise the way banks analyse customer data, identify patterns and make data-driven decisions. Leveraging these technologies could 4 aid RMs in gaining deeper insights into customer preferences, behaviours and credit risk profiles, enabling them to tailor lending products and services to individual customer needs (Sharma et al., 2023). Figure 2 below, extracted from HES Fintech’s report on Digital Transformation in Banking and Finance (2020), illustrates how digital advancements lead to enhanced customer service through improved response times. In today's competitive market, banks that do not embrace AI technologies in their lending operations, risk lagging behind and failing to meet the changing needs of their customers (Challoumis, 2024). Figure 2: Customer Journey Shift Source: (HESFintech, 2020) According to a study conducted by Gyau et al (2024), banks that leverage AI technologies in their lending operations can significantly improve customer satisfaction, reduce costs, and increase revenue. Autonomous, an international financial research firm, predicted that AI could help banks save up to $1 trillion in operating expenses globally by 2030 (Abbott, 2024; Joyce, 2024). Abbott elaborated, stating, “our models show that by pairing AI with people to offer personalised wealth advisory, guide commercial relationship conversations, tailor products for individual customers, enhance the quality of contact centre 5 interactions, and streamline their product application and onboarding processes, banks can improve their revenue by 6% or more within three years” (2024, p. 9). Despite the potential of GenAI technologies to transform lending operations, traditional banks have been slow to adopt. Reluctance can be attributed to regulatory concerns, data privacy issues, and a resistance to change (Oracle, 2021). According to Purdy and Daugherty (2017) only 5% of banks have successfully integrated AI across their organisations, with many citing challenges related to implementation and existing system integration. While existing literature acknowledges the potential of GenAI to improve banking processes, there is a clear lack of understanding regarding its integration into relationship management roles to enhance the lending customer experience. Studies by Zegullaj et al. (2023), Parmar (2023) and Hamzah et al. (2016) have all explored AI's impact on RMs ability to improve overall bank customer experience. This gap surfaced the need for empirical research, with practical implementation strategies, to bridge theoretical knowledge presented in the literature. This study aims to fill this gap by investigating how GenAI can enhance RMs ability to deliver personalised and efficient lending solutions, with actionable insights, to optimise the lending customer experience. 1.4 Research objectives The research objectives of this study to bridge the gap identified in the literature are as follows: 1. To evaluate how GenAI transforms credit lending processes and enables product personalisation in South African banks. 2. To explore how the role of RMs evolves with the adoption of GenAI and its effects on customer experience. 3. To identify the opportunities and challenges of integrating GenAI into relationship management within the banking sector. 6 1.5 Rationale The interest in undertaking this research is informed by a critical issue in the banking industry - the need for increased efficiency, accuracy, and personalisation in lending processes to enhance customer satisfaction (Ionascu & Barbu, 2023; ScienceTimes, 2024; Sharma et al., 2023). By exploring how GenAI can be integrated into relationship management roles, this research has the potential to offer practical solutions for banks to improve their lending customer experience and stay competitive in a rapidly evolving financial landscape. Investigating the impact of GenAI on RMs and customer experience in bank lending is relevant because it directly addresses the challenges of manual processes and outdated systems, faced by traditional banks, and the opportunity for GenAI to solve for this (Ek & Arnarp, 2024). By understanding how AI can enhance credit assessment, product personalisation, and customer interactions, banks can leverage technology to streamline their processes, reduce costs, and better meet their customers’ needs. This research addresses the following gaps in existing literature: the lack of empirical studies on the integration of Gen AI into relationship management roles, limited understanding of the evolving role of RMs in the AI era, and the scarcity of research on the practical challenges and opportunities of implementing AI technologies in the bank lending sector. In its attempt to solve for these gaps, this research provides valuable insights into how banks can effectively leverage AI to improve their lending customer experience. This study contributes to academic knowledge by advancing understanding of how GenAI can transform relationship management roles, enhance theoretical frameworks for integrating AI into the banking sector, and offer empirical evidence on the benefits and challenges of AI adoption. Practically, this research provides banks with a deeper understanding of the implications of GenAI for relationship 7 management practices, enabling them to make informed decisions regarding technology adoption and drive organisational strategy for improved business performance in a digitally driven economy. Policymakers may leverage the findings of this study to formulate regulatory frameworks that promote responsible AI adoption in the banking sector. 1.6 Delimitations of the study The delimitations listed below serve to clarify the boundaries and focus of this research study: i. Scope of banking sectors: The research study focuses specifically on the lending sectors within banking institutions. Other sectors within the broader banking industry, e.g., retail or investment banking, have been excluded from the study. ii. Organisational type: The study targets traditional banking institutions and excludes non-bank financial institutions like credit unions or fintech startups. iii. Level of employees: The research focuses on RMs and their interactions with customers. Other levels of employees within banking institutions, such as executive management or frontline staff, are not the primary focus of the study. iv. Conceptual delimitations: The study delimits the exploration of GenAI's impact to relationship management practices within lending sectors, excluding broader applications of AI within banking operations unrelated to customer experience enhancement. Additionally, the study does not delve into individual customer behaviours or preferences, but instead focuses on the overarching trends and implications for relationship management strategies. 8 1.7 Definition of terms I. Artificial Intelligence: Artificial intelligence, also known as AI, is a type of technology that allows computers and machines to mimic human intelligence and skills in problem-solving (IBM, 2024). II. Customer Experience (CX): Customer experience refers to the overall perception a customer has of a brand or company based on all interactions and touchpoints throughout the customer journey. It encompasses the customer's feelings, emotions, and opinions formed from various interactions with the company's products, services, and employees. Enhancing customer experience involves improving satisfaction, loyalty, and advocacy (McKinsey, 2021). III. Generative AI: Generative AI or GenAI refers to a subset of artificial intelligence that focuses on creating new or novel content or data that resembles existing data. This can include generating text, images, music, or other forms of media. Generative AI systems use machine learning techniques, often based on models like Generative Adversarial Networks (GANs) or transformer architectures, to produce outputs that are coherent and contextually relevant (McKinsey, 2023). IV. Lending Services: Lending services refer to the financial products and activities offered by banks and other financial institutions that involve providing funds to borrowers with the expectation of repayment, usually with interest. Lending services include various types of loans such as personal loans, home loans, business loans, and credit card loans (Razorpay, 2023). V. Machine Learning: Machine learning (ML) is a subset of AI that focuses on creating algorithms and models that allow computers to learn from data and make decisions or predictions without the need for specific programming for each task. ML algorithms analyse large datasets to identify patterns, relationships, and trends, and use this information to make predictions or decisions. Common ML techniques include 9 supervised learning, unsupervised learning, and reinforcement learning (Oracle, 2024). VI. Natural Language Processing: Natural Language Processing (NLP) is a branch of artificial intelligence (AI) and computational linguistics concerned with the interaction between computers and human (natural) languages. It involves the development of algorithms and techniques that enable computers to understand, interpret, and generate human language data in both written and spoken forms. NLP enables machines to analyse, process, and derive meaning from large volumes of natural language data, enabling applications such as language translation, sentiment analysis, and text summarisation (SAP, 2024). VII. Relationship Manager: A Relationship Manager (RM) is a professional employed by a bank or financial institution who is responsible for managing relationships with clients, typically high net-worth individuals, or business clients. Their role involves understanding clients' financial needs, offering appropriate banking products and services, and providing personalised advice and support (Investopedia, 2022). 1.8 Assumptions The assumptions below can fundamentally shape the research design, data collection methods, and interpretation of findings. Acknowledging their reasonableness and sensitivity can help reduce biases and strengthen the reliability of the research results. i. RMs possess a basic understanding of GenAI and its applications within the banking industry. It is reasonable to assume that RMs have a foundational knowledge of emerging technologies within their field. However, the extent of their understanding may vary and could potentially influence their perceptions and responses during interviews or surveys. The outcome of the research 10 is sensitive to this assumption as RMs with limited knowledge of GenAI may provide less meaningful insights or perspectives on its impact on customer experience. ii. GenAI adoption within banking institutions is primarily driven by a strategic focus on customer experience enhancement and operational efficiency. Given the competitive nature of the banking industry and the increasing emphasis on customer-centricity, it is reasonable to assume that all banks prioritise technologies like GenAI to improve customer experience and streamline operations. The research outcome is sensitive to this assumption as findings that contradict or challenge the assumed motivations behind AI adoption could lead to alternative interpretations of its impact on relationship management and customer experience. iii. The perspectives and experiences of RMs sampled for the study reflect standard industry practices and challenges. While it is not possible to control for every variation in individual experiences, the assumption that RMs perspectives align with broader industry standards is reasonable given their professional roles and responsibilities. The research outcome is sensitive to the assumption of unique or outlier participant perspectives that are different to industry standards as this could potentially influence the generalisation of the findings. 1.9 Conclusion The chapter introduced the study's purpose and background, setting the context for the research problem. It outlined the research objectives and reasons for conducting the study, whilst acknowledging any potential limitations. Definitions for important terms were presented, ending with underlying assumptions guiding the research. 11 CHAPTER 2. LITERATURE REVIEW AND THEORETICAL FRAMEWORK This chapter examines existing literature through three key lenses that are aligned with the research objectives: (1) GenAI's transformative impact on credit lending, (2) the evolution of RM roles, and (3) implementation challenges and opportunities. It introduces and defines the research topic and provides background relevance to the study. Following is an in-depth literature review of the objectives identified in the chapter, as well as inferred propositions that support the research problem. The chapter concludes with a discussion of the analytical framework used in the study. 2.1 Introduction The literature review begins with an examination of the transformative impact of GenAI in banking and its potential to optimise the credit lending process to provide a more personalised lending experience. Drawing on empirical evidence and industry reports, it examines the impact on enhancing operational efficiency and accelerating decision-making in traditional banking through the automation of repetitive tasks. The literature further analyses how these technology advancements reshape RM responsibilities and the impact this has on customer service. Finally, the literature examines the opportunities of integrating GenAI into the roles of RMs and the challenges that arise thereof. 2.2 Definition of topic or background discussion The process of digitalisation, known as the fourth industrial revolution (4IR), brought about significant changes in the economy (Adamek & Solarz, 2023), with almost every industry undergoing digital transformation caused by technology advancements. Krijger (2023) argues that the financial industry is no exception, 12 being early adopters to embrace technological advancements like artificial intelligence (AI) due to its potential to transform the industry. In understanding the impact of 4IR on the financial sector, it is important to first delineate between digitisation, digitalisation, and digital transformation, as these terms are often used interchangeably, yet signify distinct processes. Digitisation refers to the conversion of analog information into a digital format, essentially laying the groundwork for digital engagement (Monton, 2022). This foundational step is critical, but it alone does not produce significant change. Digitalisation, on the other hand, goes a step further, encompassing the integration of digital technologies into everyday operations, which enables businesses to improve efficiency and enhance customer interactions (Monton, 2022). This process often involves adopting tools and platforms that facilitate faster decision-making and a more streamlined customer experience. Finally, digital transformation embodies a more holistic shift, wherein organisations fundamentally rethink their strategies and business models to leverage digital technologies for new value creation (Monton, 2022). This transformation often leads to profound changes in the way businesses operate and deliver value to their customers. With these definitions in mind, it is clear that bank’s adopting AI is not merely about leveraging technology but represents a significant shift in how they operate and engage with customers, positioning them favourably in the rapidly evolving landscape of digital services. The banking sector is traditionally a very data-driven industry and, with the emergence of AI and ML technologies, there is a promise of processing larger volumes of data more effectively. This leverage could fundamentally transform important aspects of banking operations. Recognising the opportunities that AI presents to improve customer experience, streamline operations, reduce costs and make better-informed decisions, its adoption surged in recent years, making the financial industry one of the major adopters of AI (Krijger, 2023). 13 This transformation is aimed at delivering more accessible, effective and faster banking services through digital technologies (Tad et al., 2023). Sundar Pichai, CEO of Google, was quoted in (Sharma et al., 2023, p. 2) as saying, “AI is not a threat to traditional banking; it is a catalyst for its evolution”. Ionascu and Barbu attributed this to “the demands and expectations of increasingly connected and technologically savvy customers (2023, p. 55)”. Banks today offer a range of digital services, including internet and mobile banking, electronic payments, virtual assistant support and revolutionary technologies such as blockchain and AI (Ionascu & Barbu, 2023). GenAI has now brought us to the brink of a new era – one that will be shaped by automation and innovation, powered by AI (Kalia, 2023). The integration of GenAI technologies into banking services has therefore sparked due interest, with Domokos and Sajtos (2024) attributing it to the potential to boost productivity through process automation, optimisation and innovation, ultimately leading to enhanced customer experience and bottom-line revenue growth. Whilst there are the expected concerns around operational-, regulatory-, reputational- and financial risks, this research has narrowed focus to the evolving role of the RM and explores factors that could further influence adoption to enhance the lending customer experience. 2.3 Generative AI 2.3.1 Definition It is important to first understand the concept of AI before exploring GenAI. Domokos and Sajtos broadly defined AI as “a general concept that refers to the theory and development of computer systems capable of performing tasks that traditionally require human intelligence” (2024, p. 156). Donepudi elaborated, saying, “it is the process of developing intelligent computer software and systems to mimic humans by studying how humans think, how they learn, and their mental ability in solving a problem” (2017, p. 84). DBS Bank captured the meaning succinctly stating, “AI is when the computer is able to independently ‘think’ like a 14 human and perform tasks, while ML is when it ‘learns’ enough from past data or trends to carry out specific tasks and/or predict future activities” (DBS, 2023, p. 1). Several operational models such as ML, NLP and deep learning (DL) are all used interchangeably to describe AI, but each has a distinct meaning. Figure 3 below, courtesy of the Boston Consulting Group (BCG), captures these concepts and differences (Riemer et al., 2023). Figure 3: Deciphering the Jargon around AI and GenAI Source: (Riemer et al., 2023) Succinctly put, AI describes the broad concept of machines imitating human intelligence, while ML and DL are specific methods in this field (Kalia, 2023). Donepudi (2017) defines ML as a technique that allows software to analyse historical data to identify patterns and make predictions, which according to Mossavar-Rahmani and Zohuri (2023) is made possible through the availability of large training data sets. DL, a subset of ML, aims to simulate the brain's structure using layered neural networks (Emmert-Streib et al., 2020). GenAI, on the other hand, is a subset of AI which uses ML models to learn from data and create new content (Kalia, 2023; Osei et al., 2023; Tad et al., 2023). 15 Therefore, instead of relying on current sources like AI does, it is able to anticipate the next step in the pattern it identifies to create new data, images, and other forms of content (Domokos & Sajtos, 2024; Donepudi, 2017). Through training on large data sets, these algorithms progressively understand the characteristics of the various media types they are expected to produce (Krause, 2024). This knowledge allows them to generate new content in the future that is similar to the data they were trained on (Emmert-Streib et al., 2020; TechTarget, 2024). BCG (Riemer et al., 2023) observed that while many associate GenAI in banking primarily with customer service chatbots, its capabilities go well beyond that. The authors profess that GenAI can automate financial analysis and assist with code development, with numerous global banks such as Goldman Sachs, Deutsche Bank and American Express, exploring these applications - whether by developing them internally, or using services (Riemer et al., 2023). As financial leaders consider the new opportunities presented by GenAI alongside existing predictive AI solutions, it is noteworthy to recognise that AI applications can enhance nearly every aspect of bank workflows, from customer interactions to backend processes. To effectively leverage GenAI, banks will need to improve their strategies for identifying, prioritising, and nurturing initiatives that will deliver significant value for customers and employees. Big gains from AI happen when banks rethink how an entire process operates, not just the adding of AI in each stage. This means redesigning processes to consider both AI and human roles for the best results. For bank relationship management, this would mean combining GenAI with RM skills, enabling them to work more efficiently and effectively together to create much better results, rather than if they worked separately to achieve the same. 16 2.4 Transformative potential of GenAI 4IR (Mehdiabadi et al., 2020) has brought about unprecedented advancements in technology and transformed the financial sector (Khanchel, 2019). 4IR technologies, such as blockchain, Internet of Things (IoT), and big data analytics, have revolutionised the way financial institutions operate, disrupting traditional business models, and driving innovation in the industry (Shanti et al., 2022). These technologies have enabled greater efficiency, improved security, and enhanced customer experiences. The Financial sector, more specifically banking, is undergoing another significant transformation with AI being the catalyst in reshaping traditional operations (Mossavar-Rahmani & Zohuri, 2023) and heralding a new era of efficiencies and customer centric services (Tad et al., 2023). Few technologies have evolved at the pace of GenAI - or have the potential to drive such a profound reinvention. Accenture posits that “Leading banks now recognise that the question is no longer whether GenAI will transform their industry, but how?” (Accenture, 2024, p. 36). A further international study conducted by Accenture (Abbott, 2024) using data from the US bureau of labour statistics, examined 19,265 tasks in 900 job families across 19 industries, and found that the banking sector would be significantly more affected by GenAI than to other industries. 73% of all banking tasks have a potential to be impacted by AI - 39% have a higher potential for automation, and a further 34% for augmentation (Figure 4). The study concluded that banks that adopt GenAI, can increase their productivity by 30%. 17 Figure 4: The impact of GenAI on different industries Source: Accenture (Abbott, 2024) ChatGPT(2022), an AI-powered NLP tool developed by OpenAI (Sharma et al., 2023), was launched on November 30th, 2022 (Banh & Strobel) to enable human- like conversations with a chatbot capable of answering questions and aiding in tasks like composing emails, essays, and code. According to Sharma et al., OpenAI utilises Reinforcement Learning from Human Feedback (RLHF) during training, involving testers in scoring and selecting the best outcomes (2023). Key applications include improved operational efficiencies and enhanced customer engagement (Mossavar-Rahmani & Zohuri, 2023). Early adopters like Standard Bank quickly embraced the tool, collaborating with Microsoft to test MS 365 Copilot, which incorporates ChatGPT functionality (ITWeb, 2024). Group CIO, Jorg Fischer, emphasised the bank's commitment to leveraging AI language models to enhance customer service and employee performance. Now, nearly every major bank has adopted a GenAI strategy, conducting their own proofs of concept to explore the potential of AI-driven tools in their operations (Abbott, 2024; Crosman, 2024; ITWeb, 2024; Singh & Singh, 2024). 18 2.4.1 Automation potential – Credit Assessments AI and ML technologies can automate various financial processes, including customer service, loan processing, underwriting, risk assessment, and fraud detection (Tad et al., 2023). Kayode (2024) advocated that ML algorithms can automate underwriting processes through analysis of large amounts of data to determine creditworthiness, approve loans quicker and eliminate the need for manual reviews. Kalia (2023) corroborated this, adding that analysis of big data, credit history and income amongst other factors, allows ML models to generate more accurate credit risk profiles and optimise the loan approval process. The literature suggests that AI and ML technologies can automate bank lending processes leading to increased accessibility and efficiency of financial services. Sharma et al. (2023) illustrates the various steps in the lending life-cycle that GenAI can support and automate in Figure 5 below. Figure 5: Potential impact of GenAI across the lending journey Source: (Sharma et al., 2023) 19 Analysis of the lending life-cycle presented suggest many opportunities for automation: Lead generation In the current traditional lending cycle, challenges experienced by RMs with lead generation for bank lending include difficulty in identifying high-quality leads, low conversion rates, and time-consuming manual processes (Palaparthy & Pillalamarri, 2015). GenAI can alleviate these problems by automating lead scoring and streamlining the application process through chatbots (Kayode, 2024), ultimately increasing conversion rates and improving efficiency for bank RMs (Zegullaj et al., 2023). Loan application Mehdiabadi et al. (2020) pinpointed some of the challenges with the current bank loan application process to include the lengthy processing times, high rejection rates, bias in decision-making and limited RM availability for personalised guidance to applicants. GenAI can mitigate by streamlining the application process through automation of various aspects such as document gathering, verification and underwriting (Sharma et al., 2023). This can help reduce processing time and increase efficiency. GenAI can further enhance the decision- making process by analysing a broader set of data points to evaluate creditworthiness and improve accuracy in the loan approval decisions (Adamek & Solarz, 2023). This can further reduce rejection rates and ensure fair and unbiased decision-making (Oracle, 2021). Another opportunity would be for GenAI to provide personalised guidance to applicants by tailoring recommendations and advice based on their individual financial situation and needs (Singh & Singh, 2024). This can help applicants better understand their options and make informed decisions. Credit analysis Sadok et al. (2021) attributed the challenges in lending credit analysis to traditional time-consuming processes, subjectivity - where RMs or loan officers 20 introduce bias into the decision-making process, leading to inconsistent outcomes and; limited data sources. The authors went on to propose that GenAI could solve for time constraints by quickly analysing large amounts of data and providing insights into an applicant's creditworthiness in a fraction of the time it would take a RM or credit analyst. Bhatore et al. (2020) further posited that GenAI can help remove bias and subjectivity from the decision-making process by basing decisions on data-driven insights rather than human judgment. The authors suggest that GenAI can integrate data from multiple sources and formats, providing a more comprehensive and accurate view of an applicant's creditworthiness. This could lead to a streamlined credit analysis process and more informed lending decisions (Bhatore et al., 2020). The literature highlights numerous automation opportunities for the use of GenAI and suggests that banks leveraging these capabilities can enhance the overall lending experience for both customers and RMs, leading to more informed and fair lending decisions. 2.4.2 Innovation potential - Product personalisation AI innovation is reshaping the financial services landscape, as banks embrace cutting-edge technologies to streamline operations and enhance customer experiences. “AI-powered chatbots and virtual assistants enable personalised customer interactions, providing real-time support and assistance with inquiries, account management, and financial planning”, (Kayode, 2024, p. 4). An example of an innovative implementation is that of PenFed - Pentagon Credit Union (Olavsrud, 2023). PenFed utilised GenAI, with the help of Salesforce's Einstein platform (Parmar, 2023), to revolutionise its customer interactions and significantly improve service efficiency and member satisfaction. This was achieved by deploying chatbots (Kaliuta, 2023) internally and externally. PenFed’s AI journey has yielded impressive results, with a 223% increase in chat and chatbot activity in the past year and a 20% resolution rate on first contact. In 21 a South African context, the opportunities are vast for implementing similar AI technologies for bank lending purposes. By leveraging GenAI like PenFed has done, banks can enhance customer engagements, create more efficient service delivery (Tad et al., 2023), and improve customer satisfaction (Singh & Singh, 2024). AI-powered chatbots can streamline loan application processes, provide real-time support to customers (Kaliuta, 2023), and offer personalised assistance based on individual needs. Additionally, AI can be used to gather data insights (Dhake et al., 2024) and analyse customer behaviour, allowing banks to offer more tailored lending solutions (Sharma et al., 2023). By leveraging AI to deliver exceptional value to customers (Singh & Singh, 2024), South African banks can stay ahead of the competition and better serve their customers in an increasingly digital world. The literature suggests that there is significant potential for AI innovation in lending, especially for commercial and retail banks. An opportunity presented by Ooi et al. (2023) was for banks to consider developing product or domain-focused GenAI tools to enhance their business operations. This holds huge potential for this research study. What the authors suggest, is that the product or domain of lending services has a significant amount of data sources and can start training ML models with these data sources to develop and provide a platform for personalised consumer engagement. While GenAI is still in its infancy in bank lending sectors, early adopters will stand to benefit significantly in terms of product personalisation, customer satisfaction, and overall business success. 2.5 Relationship management The banking sector’s lending success relies heavily on the individual roles of RMs (Colgate & Lang, 2005). Zegullaj et al. (2023) suggested that effective relationship management is crucial for businesses and organisations to build loyalty, trust, and long-term partnerships with their customers and clients. 22 Therefore, by nurturing these relationships, banks can increase customer satisfaction, customer retention, and ultimately, their bottom line. In a study by Hamzah et al. (2016) determining the role of the RM on the impact between customer behaviour and market performance, it was concluded that RMs often find themselves pitted against conflicting expectations; (1) the need to generate revenue for their businesses through sales or utilisation of banking facilities, (2) protecting the bank's interests through due diligence and risk management, and (3) proactively meeting their customers’ needs and making sound judgments pertaining to risk and return. 2.5.1 Evolving role in the era of AI – Impact on customer experience The rapid advancement of digitisation in the banking industry has greatly impacted the traditional role of RMs in bank lending (Colgate & Lang, 2005). As banks continue to embrace technology and automate various processes, RMs are tasked with adapting to new responsibilities and roles to enhance customer experience (Adamek & Solarz, 2023). This shift in focus towards digital solutions has, not only changed the way RMs interact with customers, but also significantly impacted the overall lending customer experience (ScienceTimes, 2024; Sharma et al., 2023). Digitisation has influenced the role of RMs by providing them with advanced tools and technologies to better understand and deliver to their customers' needs (Adamek & Solarz, 2023; HESFintech, 2020). Previously, these advisors spent considerable time manually processing and analysing customer data, relying on traditional Customer Relationship Management (CRM) systems and spreadsheets (Ionascu & Barbu, 2023). GenAI now automates data aggregation, analysis, and insight generation (Dhake et al., 2024; Tad et al., 2023), allowing RMs to access comprehensive client profiles instantly and focus on strategic activities such as personalised client services. This is made possible through 23 analysis of large amounts of data (Zegullaj et al., 2023) and financial planning (Domokos & Sajtos, 2024). Raj Abrol, CEO of Galytix (2023), a GenAI data technology company, stated in a press release that the company firmly believed in leveraging AI capabilities to assist RMs and credit analysts in banks by automating routine tasks to enhance productivity. “It's like having an analyst on demand" (Galytix, 2023). Thus, by leveraging data analytics and AI, RMs can track customer behaviour, preferences, and transaction history to offer targeted products and services. Ooi et al. (2023) provided the example of a RM inputting a customer’s transactional history into a GenAI model to understand the spending habits, and proposed the offering of a new savings account or product as a personalised service to the customer. In addition, the authors suggested that GenAI models can be used for training RMs, streamlining repetitive and time-consuming tasks, serving as a virtual assistant to respond to customers’ increasingly complex inquiries, as well as releasing staff from routine tasks. The authors proposed that this would allow for more focus on innovative and strategic responsibilities which would lead to increased productivity, efficiency, and accuracy (2023). In an article that examined how GenAI can add value in banking and financial service, Gökhan Sari (2023), head of McKinsey's Financial Services and Risk Practices in Africa, Eastern Europe and the Middle East, suggested using a copilot to assist RMs in delivering improved service. Rather than spending time pouring over vast amounts of information before meeting with a client, the copilot utilises AI technology to analyse client data and provide recommendations quickly. This would allow for more focused and effective client discussions (McKinsey, 2023). Other use cases that banks are testing include the use of GenAI based virtual assistants to help small-business owners through the loan application process (Chen et al., 2023; Khanchel, 2019), as in the case of Bankwell Bank in Connecticut (Crosman, 2024). The Bank partnered with Cascading AI to streamline the small business loan application process (Crosman, 2024). The 24 company’s Casca AI software acts as a virtual assistant named Sarah, to prequalify loans, thus handling the time-consuming, tedious tasks such as data validation and loan approval. This reduced the need for Bankwell’s RMs to spend hours guiding small business lenders through the application process. Sarah runs pre-qualifying tasks with potential lenders 24/7, allowing for faster and more efficient loan approvals. This technology helped to improve the loan application process, resulting in time savings and a more positive overall experience. Bankwell’s case study suggests that GenAI has the potential to revolutionise the way in which RMs respond to customer request and needs. Tad et al. (2023) corroborated, stating that GenAI can augment RM capabilities with advanced automation, particularly in complex and resource-intensive processes like small business lending (ScienceTimes, 2024). Additionally, AI's ability to operate autonomously around the clock (Ali & Aysan, 2023), as seen with Sarah handling tasks at late hours, surfaced its role in providing timely and responsive service, which could ultimately improve customer satisfaction and retention (Zegullaj et al., 2023). The literature reviewed suggest that GenAI is reshaping the role of the RM in banking, from transaction processors to trusted advisors, who leverage technology to provide better customer experience and satisfaction (Mossavar- Rahmani & Zohuri, 2023; Zegullaj et al., 2023). This includes an understanding of client preferences, financial goals and risk tolerance, allowing them to suggest tailored financial products or investment strategies, thereby offering an elevated client experience. 2.6 GenAI in relationship management GenAI has exponentiated the potential to revolutionise the way RMs operate within the banking industry (Dhake et al., 2024; Parmar, 2023). Donepudi (2017) suggests that by leveraging ML algorithms these AI systems can analyse large amounts of customer data in real-time and provide personalised 25 recommendations and insights (Bhatore et al., 2020). This can lead to more proactive and tailored interactions with customers, improve satisfaction and loyalty (Zegullaj et al., 2023) and ultimately increase banks competitive edge in the market (Abbott, 2024). Conversely, Singh (2024) claims that integrating GenAI into RM practices can also present its fair share of challenges, such as concerns over data privacy, ethical considerations, and the need for upskilling and training RMs to effectively utilise these technologies. This section explores the opportunities and challenges of incorporating GenAI into the role of RMs at banks. 2.6.1 Opportunities 2.6.1.1 Personalisation at Scale GenAI and ML algorithms, not only enable personalised customer experiences by analysing customer data and behaviour (Riemer et al., 2023), they also empower RMs to utilise this data and behaviour insights to tailor financial products and services to each client's unique needs. By leveraging these advanced technologies (Parmar, 2023), RMs can strengthen their customer relationships, increase engagement, and ultimately enhance customer satisfaction. Additionally, AI-powered robo-advisors provide automated investment advice (Tad et al., 2023), allowing RMs to dedicate their attention to providing tailored and personalised service to clients with more complex financial needs. Overall, the combination of personalised recommendations, automated services, and efficient mobile apps, provided by banks who are equipped with these digital enhancements (Olavsrud, 2023), allows RMs to better serve their clients and deliver exceptional customer convenience (Kayode, 2024). 2.6.1.2 Predictive Analytics Predictive AI is a type of AI that assists banks with predicting and classifying risks, determining optimal pricing, and modelling product preferences (Riemer et 26 al., 2023). GenAI and predictive AI serve different purposes, with predictive AI focusing on logic and calculation while GenAI excels in creativity and expression (Devan et al., 2024). Riemer elaborated by saying that both are important for a bank's AI strategy as they complement each other like the two halves of the human brain. Predictive AI, like the left brain, handles probabilities and decisions, while GenAI, likened to the right brain, is skilled at generating human-like responses in automated chats (Riemer et al., 2023). In order to harness the strengths of both types of AI, a bank's strategy should include both predictive and GenAI with RMs leveraging both to anticipate customer needs and behaviours based on historical data, predict issues or opportunities, and then offer proactive solutions. 2.6.1.3 Intelligence Automation Automation has been proven to increase productivity and efficiency (Singh & Ahuja, 2024). Robotic Process Automation (RPA), a feature of AI, can automate routine tasks such as data entry and report generation, thus enabling banks to reduce expenditure, simplify operations and free up RMs to focus on strategic activities such as client engagement and problem-solving (Romao et al., 2019). Through leverage of RPA, RMs can focus on building stronger relationships with their clients and providing more personalised and efficient service (Romao et al., 2019). 2.6.1.4 24/7 Availability Traditional banking processes involve the manual sorting and analysing of massive amounts of paperwork by RMs, resulting in wasted time and incurring additional costs (Palaparthy & Pillalamarri, 2015). AI-driven chatbots or virtual assistants can provide round-the-clock support (Kayode, 2024), ensuring that customers receive assistance whenever needed, thus improving overall service availability, as in the case study of Bankwell Bank (Crosman, 2024). GenAI, having its expertise in dealing with large data sets, can process large volume of 27 documents , identify important data and summarise this at a fraction of the time and cost it would take a RM (Singh, 2024). 2.6.2 Challenges 2.6.2.1 Algorithmic Bias The algorithms that power AI services can also introduce complex ethical challenges and one of the most prevalent is the potential to inherit and amplify existing biases (Oracle, 2021). GenAI systems learn from historical data, and should the existing data contain any biases, the system could exacerbate these (Domokos & Sajtos, 2024). From a banking perspective, this could result in prejudiced decision-making in critical financial services, such as lending, insurance, and investment. For instance, if an AI system is trained using biased historical lending information, it could end up unfairly rejecting loan applications from a specific demographic group, thereby reinforcing existing inequities (Matsie, 2023; Oracle, 2021). 2.6.2.2 Skill Gaps and Training The way in which banks work is about to change fundamentally. According to Singh (2024), innovation is happening globally by skilled people who know how to use GenAI technologies, and who can solve problems using them. E-learning company Coursera claimed that 35 GenAI courses offered by global universities had an intake of 196000 enrolments in India alone in 2023 (Singh, 2024). A study conducted by Accenture (2024) posits that new skills, approaches and mindsets will be needed in every function and level of the bank. The study indicates that recruitment alone cannot solve for the challenges and that a completely new approach to reskilling is required. In context of relationship management roles, this may include upskilling for effective use and interpretation of AI-generated insights. Ooi et al. (2023) highlighted a key issue - while GenAI is evolving at a breakneck pace, there is still a lack of understanding as to how 28 such technologies can be integrated into the workplace in a way that generates value. The authors differentiated 2 groups of employees; those that are familiar with GenAI and see the value in utilisation and, those that have little knowledge or are apprehensive about using such technologies (Ooi et al., 2023). These distinct divisions may result in conflicts and disagreements, especially in the lending industry, where RMs, who are accustomed to traditional methods (Colgate & Lang, 2005; Ooi et al., 2023), may resist new changes and be hesitant to adopt them, resulting in delays in implementation and possible pushback. 2.6.2.3 Customer Acceptance While automation has its benefits, there have been unwelcome side-effects (Domokos & Sajtos, 2024). According to Accenture’s Banking Top 10 Trends for 2024, “by shifting customer engagement out of the branch and onto their digital channels, the banking experience has become functionally correct but emotionally void due to the reduced need for human interaction” (Abbott, 2024, p. 11). As banks’ individual and personal connection with customers wane, so too does their ability to differentiate themselves from their competitors (Domokos & Sajtos, 2024), thus making the job market that much smaller (Sharma et al., 2023). This is not to discount that some customers may prefer human interaction over AI-driven services, which raises concerns about trust and the personal touch typically associated with traditional banking relationships (Colgate & Lang, 2005) and questions brand loyalty. On average, customers have 6.3 financial products, with only half of them coming from their main bank. 73% of customers have obtained at least one new financial product from a different provider within the last year. (Abbott, 2024). 2.6.2.4 Regulatory Impact The banking sector is a highly regulated environment where protection of consumer data, adherence to privacy laws, and the maintenance of ethical standards are paramount (BCG, 2023). Banks must address data privacy and 29 protection, maintain security and ensure compliance and accountability. Each of these areas requires careful consideration and adherence to relevant regulations and best practices. Data privacy and protection: GenAI systems require access to large amounts of customer data to learn and train effectively. This data often includes sensitive information such as financial transactions, personal identification details, and communication history. Ensuring that the data is handled in compliance with data protection regulations is a significant challenge, which includes obtaining explicit consent from customers for data usage and ensuring robust data security measures (Krijger, 2023). Security risks: Integrating AI into banking operations can heighten cybersecurity risks. GenAI models require robust infrastructure and effective data management, which can complicate a bank's cybersecurity landscape. Financial institutions face stringent cybersecurity regulations that mandate rigorous security measures (Kalia, 2023). Banks must address the implications of AI on their cybersecurity practices. Vulnerabilities in AI systems can lead to data breaches or financial fraud, potentially violating regulatory requirements and undermining customer trust. To mitigate these risks, banks should implement strict security measures, including regular security audits, data encryption, and protection against adversarial attacks. Adhering to regulatory frameworks and specific security standards is essential to safeguarding both the AI systems and the sensitive data they handle. Compliance and Accountability: Financial regulations demand detailed documentation and accountability for decisions made by financial institutions. When AI systems are used in relationship management, it's crucial to have clear accountability for AI-driven decisions that impact consumers (Abbott, 2024). Banks need to ensure they can audit these decisions, if necessary, which involves maintaining logs of AI interactions and providing explanations and justifications for AI-generated outcomes. Regulators are particularly concerned 30 with identifying responsible parties when AI systems make errors that result in financial losses or breaches of consumer trust. This accountability is essential for maintaining transparency and addressing any issues that arise from AI-driven decisions. 2.7 Analytical framework The banking sector is undergoing major digital transformation, with GenAI disrupting traditional banking (Ionascu & Barbu, 2023; Kalia, 2023; Khanchel, 2019). This presents a persuasive opportunity for South African banks to optimise customer service and elevate the customer experience through automation and innovation. This research leverages existing theoretical models for a worldview synthesis of information technology (IT) adoption to understand the factors that influence the adoption of GenAI technologies by RMs in bank lending services. 2.7.1 Theoretical Framework Various technology adoption models were researched, both at an individual and an organisational level. 2.7.1.1 The Technology Organisation Environment (TOE) The TOE framework was considered at an organisational level as it focuses on the interactions between technology, organisation, and the external environment in determining technological adoption and implementation (Na et al., 2022). It considers factors such as technological readiness, organisational structure, and external market conditions. TOE, in context of this research, would be the adoption of GenAI (technology) within the bank lending sector (organisation). External environment would refer to the different customer segments, and regulatory and ethical considerations, for the use of GenAI. Factors such as ease of use, perceived benefits and usefulness, complexity of integration, and IT infrastructure are usually considered relevant in influencing the IT adoption 31 process (Huy et al., 2024). These factors are all pertinent to the adoption of GenAI in banking. It is for this reason that the TOE framework has been a proven theory for the study of digital innovation adoption at organisation level and has been validated through several studies (Huy et al., 2024; Na et al., 2022). The limitation for adopting the TOE framework for purposes of this research, is that it does not consider the individual in this worldview, i.e., the RM, and would therefore not fulfil the complete research objective. It is for this reason that the model does not suffice for purposes of this research. 2.7.1.2 The Technology Acceptance Model (TAM) Another theoretical model considered was TAM, which focuses on the individual’s perception and attitude toward technological adoption (Na et al., 2022). TAM is a widely-used theoretical framework and has been applied in various studies, (Dwivedi et al., 2017; Na et al., 2022; Venkatesh, 2022), that sought to explain users' acceptance and usage of technology. It emphasises perceived usefulness and perceived ease of use as key determinants of technology adoption (Dwivedi et al., 2017). Perceived usefulness is the belief that a person has that a particular system would improve their job performance. In the context of this research, TAM can be applied to understand how RMs perceive the usefulness and ease of use of GenAI technology in their role (Hamzah et al., 2016), and how this perception influences their intention to integrate AI into their practices. This framework can help identify potential barriers to adoption and implementation of GenAI in relationship management, as well as factors that might facilitate its successful integration (Domokos & Sajtos, 2024). The limitation with TAM though, is that it does not consider the organisation being impacted, i.e., the bank lending sector, and therefore falls short of fulfilling the theoretical worldview required here. It is for this reason that TAM is not used in this research. 32 2.7.1.3 Unified Theory of Acceptance and Use of Technology 2 (UTAUT2) Unified Theory of Acceptance and Use of Technology 2 or UTAUT2, is a model that helps explain how, and why, people decide to use new technology. It builds on the original UTAUT model of Performance Expectancy, Effort Expectancy, Social Influence, and Facilitating Conditions, and adds 3 more factors – Hedonic Motivation, Price Expectancy and Habit (Dwivedi et al., 2017; Venkatesh et al., 2012), as shown in Figure 6 below. Figure 6: The unified theory of acceptance and use of technology2 (UTAUT2) Source: (Venkatesh et al., 2012) Performance Expectancy: This is about how much a person believes that using the technology will help them perform better or achieve their goals (Venkatesh, 2022). Effort Expectancy: Refers to how easy or difficult a person thinks it will be to use the technology. The simpler it is to use, the more likely they are to adopt it (Dwivedi et al., 2017). Social Influence: Considers how a person is influenced by others around them (like friends, family, or colleagues) when deciding to use a technology. If 33 important people in their lives support the use, they are more likely to use it (Venkatesh, 2022). Facilitating Conditions: Involves the resources and support that are available to help someone use the technology. Things like training, availability of hardware, or access to the internet are included here (Ayeni et al., 2024). Hedonic Motivation: This is about the enjoyment or pleasure a person gets from using the technology. If it’s fun or enjoyable, they are more likely to use it (Dwivedi et al., 2017). Price Value: Considers the cost of using the technology versus the benefits it provides. If the benefits outweigh the costs, people will be more inclined to adopt it (Venkatesh, 2022). Habit: Refers to the extent to which people automatically engage in using technology as a result of prior actions or experiences. Habit is a critical aspect as, once users become accustomed to using technology, the likelihood of continued use increases even if performance or effort factors fluctuate (Venkatesh et al., 2012). Behavioural Intention: Pertains to a user’s intention to engage with the technology, which is influenced by the other factors in the model. Behavioural intention serves as a direct predictor of actual technology use, as it indicates the user's conscious decision to adopt a particular technology (Dwivedi et al., 2017). Use Behaviour: reflects the actual usage of the technology, capturing how different users engage with the technology in practice. This would help determine if the expected benefits from the adoption of a technology are realised and provide insights into user interactions after the technology has been integrated into the business (Venkatesh et al., 2012). In essence, the comprehensive nature of the UTAUT2 model, encompassing not only the initial acceptance of technology but also the habitual usage patterns and the transition from intention to actual behaviour, makes it ideal for evaluating the multifaceted impacts of GenAI on customer experience and relationship management in the banking sector. The framework is particularly applicable 34 where technology adoption is influenced by more diverse and nuanced factors such as acquisition of new skill sets, resistance to change and ethical bias associated with the evolving role of RMs in the era of GenAI The research objectives of the study align closely with the UTAUT2 model for several reasons: Evaluating Transformation: The model facilitates understanding the complex interplay between technology and users' perceptions in transforming credit lending processes, focusing on key factors like Performance Expectancy and Effort Expectancy, which are central to adopting GenAI. Evolving Roles of RMs: By incorporating behavioural factors like Habit and Behavioural Intention, UTAUT2 supports examining how RMs integrate GenAI into their practices, helping reveal shifts in professional roles and customer engagement strategies. Opportunities and Challenges: The model's comprehensive framework allows for a nuanced exploration of barriers e.g., resistance to change, and facilitators of GenAI adoption in relationship management, aligning with the objective of uncovering opportunities and challenges within this context. Overall, UTAUT2 serves as a robust framework that provides both theoretical grounding and practical insights for this study, focusing on technology acceptance and usage within the evolving banking landscape, thus justifying its selection over other models like TAM and TOE. Again, whilst TAM and TOE are acceptable models, they are limited in that TAM primarily focuses on individual users' acceptance of technology, while TOE is more centred on the organisational context of technology adoption (Na et al., 2022). This study therefore has adopted UTAUT2 as the theoretical framework within which this research is conducted. 35 2.7.2 Conceptual Framework The literature review conducted considered some of the most applied theories used in technology adoption, i.e., TOE, TAM and UTAUT2 (Dwivedi et al., 2017; Na et al., 2022). The conceptual framework of this study is based on the UTAUT2 theoretical model (Venkatesh, 2022; Venkatesh et al., 2012). The UTAUT2 model, which identifies performance expectancy, effort expectancy, social influence, and facilitating conditions as some of the critical determinants of technology adoption (Venkatesh, 2022), is particularly relevant to this research. In this context, the adoption of GenAI by RMs is shaped by more nuanced factors, including the evolution of their roles, the need to acquire new skill sets, and their ability to adapt to change. The study therefore seeks to determine whether GenAI plays a mediating or moderating role in influencing the customer experience provided by RMs. The causal loop diagram, used to set intent for undertaking this research in chapter 1, is referenced again in Figure 7 below as the conceptual framework against which the research problem is explored. Figure 7: Causal Loop Diagram Source: Author’s own. The UTAUT2 adoption theory has been mapped against the 3 key variables in the causal loop reinforcement diagram and seeks to establish whether GenAI could effectively influence RMs’ ability to enhance the lending customer 36 experience. The research attempts to prove that augmented GenAI technology could boost the skills of RMs, which in turn could lead to better customer experiences. The research also aims to validate the feedback loop theory by aligning the main propositions surfaced in this study within the UTAUT2 theoretical framework, which indicates that customer feedback can enhance GenAI systems. The consistency table presented in Table 1 below aligns each research objective to its mapped proposition, as well as the relevant data collection and analysis methodology utilised in the study. RO # Research Objective Proposition Data collection detail Data analysis method 1 To qualitatively analyse the impact of GenAI on the credit lending process and product personalisation in bank lending P1: Through automation and innovative adoption of GenAI technologies, banks can revolutionise their credit lending processes and product personalisation offerings to improve customer satisfaction and drive economic growth. Interview guide questions (Appendix A1 and Appendix A2) Thematic analysis 2 To explore how the role of RMs evolves with the adoption of GenAI and its effects on customer experience. P2: Adopting GenAI transforms the role of RMs from transaction processors to trusted advisors, significantly enhancing customer experience through personalised and strategic interactions. Interview guide questions (Appendix A1 and Appendix A2) Thematic analysis 3 To identify the opportunities and challenges of integrating GenAI into relationship management within the banking sector. P3: GenAI adoption in bank lending services presents both opportunities and challenges which must be overcome for successful integration. Interview guide questions (Appendix A1 and Appendix A2) Thematic analysis Table 1: Consistency table Source: Authors own 37 2.7.2.1 Proposition 1 (P1) Through automation and innovative adoption of GenAI technologies, banks can revolutionise their credit lending processes and product personalisation offerings to improve customer satisfaction and drive economic growth. This proposition suggests that banks can enhance the efficiency and personalisation of their services by incorporating automation and GenAI technologies into their credit lending processes and product customisation strategies. This, in turn, can lead to higher levels of customer satisfaction and ultimately contribute to economic growth. The ways in which GenAI can potentially enhance the efficiency and personalisation of credit lending processes are listed as follows: Automated credit scoring: GenAI can streamline the credit scoring process by analysing vast amounts of data to accurately assess creditworthiness. This automation can reduce processing times, minimise errors, and improve the efficiency of loan approvals. Personalised product recommendations: By analysing customer data and behaviour patterns, GenAI can offer bespoke product offerings tailored to individual financial needs and preferences. This level of customisation can enhance the customer experience and drive customer satisfaction. Real-time decision-making: GenAI can enable real-time loan approvals or rejections based on advanced algorithms that assess risk factors and predict credit outcomes. This can speed up the decision-making process, leading to faster loan processing and improved customer service. Empirical evidence supporting the above include the research conducted by Sharma et al. (2023) on the benefits of GenAI in assessing loan eligibility in the banking sector. The study demonstrated how AI models can accurately analyse customer data, evaluate potential risks, and provide real-time loan evaluations, 38 leading to more informed lending decisions and improved efficiency. Additionally, Singh and Singh (2024) found that AI chatbots, powered by GenAI models, can handle customer enquiries, offer tailored product recommendations, and assist with banking transactions. This can lead to higher levels of customer satisfaction and loyalty. Brown et al. (2020), in their study, highlighted the benefits of NLP, including ChatGPT, in enhancing customer engagement, streamlining processes, and improving the overall customer experience in banking. Furthermore, case studies like Bankwell Bank (2024) and Valley Bank (2024) noted the improvements in efficiency, accuracy, and customer satisfaction through the use of AI-powered algorithms for credit scoring and personalised product recommendations. Figure 8 below illustrates the findings from a joint study conducted by Microsoft Asia and IDC Asia-Pacific, as reported in the Bangkok Post (2019). The results reveal significant improvements reported by financial services organisations that have already begun adopting AI. These include higher margins, better business intelligence, accelerated innovation, better customer engagement and higher competitiveness - recorded in a range of 17% to 26% improvement. This was expected to accelerate to between 35% and 45% after 3 years, an estimated increase of 2.1 times. South African banks have been slower in adoption but there is an expectation to see similar possible trends with the inclusion of GenAI in banking services. 39 Figure 8: Future Ready Business Source: (BangkokPost, 2019) The proposal aligns with the UTAUT2 principles of performance expectancy, effort expectancy, and social influence constructs (Venkatesh, 2022), suggesting that integrating GenAI into credit lending and product personalisation processes, through the use of AI-powered chatbots in banking can optimise operations (performance expectancy), provide tailored recommendations (effort expectancy), and enhance customer experience (social influence). Oracle (2021), however, argued that integrating GenAI into bank lending services could prove to be challenging. The example given was of algorithmic biases in data, resulting in unfair lending practices, which would need to be addressed. Banks would need to implement strategies that guarantee transparency, accountability and fairness in their AI decision-making processes to circumvent this. From a broader economic perspective, the proposition can have positive impacts on access to credit for SMEs. Leveraging GenAI to streamline their lending process would enable banks can make it easier and faster for small businesses to obtain loans (Crosman, 2024). This can stimulate economic growth, encourage entrepreneurship, and support job creation in the SME sector. By adopting GenAI technologies, banks can differentiate themselves in a competitive market, attract 40 and retain customers, and drive economic growth through more targeted and efficient lending practices. Challenges relating to the proposition include concerns about data privacy, regulatory compliance, and the potential displacement of human workers. These, however, could be addressed through robust data protection measures (Dhake et al., 2024), adherence to regulatory guidelines, and upskilling of employees to work alongside AI technologies (Zegullaj et al., 2023). The reliability and accuracy of AI algorithms came into question, but the literature argues that advancements in AI technology and ongoing monitoring and evaluation can mitigate these concerns (Kaliuta, 2023). The proposition that banks can revolutionise their credit lending processes and enhance customer experiences through the adoption of GenAI technologies is supported by existing research, case studies, and practical implications. By leveraging AI to automate processes, personalise products, and empower RMs, banks can drive growth, improve customer satisfaction, and stay competitive in a rapidly evolving industry. 2.7.2.2 Proposition 2 (P2) Adopting GenAI transforms the role of RMs from transaction processors to trusted advisors, significantly enhancing customer experience through personalised and strategic interactions. The adoption of GenAI will be driven by the need for a more customer-centric approach in a digitalised banking environment (Zegullaj et al., 2023). This shift in role from transaction processor to trusted advisor can enable RMs to provide an enhanced customer experience in multiple ways: Personalised and proactive recommendations: GenAI will enable RMs to analyse large amounts of data and identify patterns to provide tailored advice to clients (Devan et al., 2024). For example, AI algorithms can analyse customer transaction history, spending patterns, and financial goals to offer tailored 41 recommendations on investment opportunities or loan products to bank lending customers. Streamlined processes: GenAI can automate routine transaction processing tasks, allowing RMs to focus on strategic advisory functions. For instance, AI- powered chatbots can handle customer queries (Kayode, 2024), freeing up RMs time to engage in more valuable interactions with clients. Enhanced customer experience: By leveraging GenAI technology, RMs can offer more customised solutions and improve overall customer experience (Hamzah et al., 2016). For example, AI-powered tools can predict customer needs and preferences, enabling RMs to anticipate client requirements and provide proactive support. The practical implications of this proposition are significant for financial institutions and their RMs. By adopting GenAI technology, RMs will have access to robust data analysis tools that can help them better understand their customers' needs and preferences (Hamzah et al., 2016). This will allow for more personalised advice and solutions, ultimately leading to better outcomes for the customer. Empirical evidence supporting this proposition include a study by McKinsey (2023), highlighting how AI can enhance the capabilities of RMs by enabling them to provide bespoke advice to clients. This is accomplished through analysis of large amounts of data and identifying patterns leading to tailored solutions. Furthermore, a recent report by Deloitte (2024) on GenAI and wealth management, discussed how GenAI technology can assist wealth RMs in better understanding their clients' individual needs and preferences, enabling them to offer more customised and valuable recommendations. “Digital solutions augment rather than replace the human element which remains at the core of the industry’s value proposition” (Deloitte, 2024). Other case studies include those of HSBC (White, 2021) and DBS Bank (2023), who both implemented AI-powered solutions where routine transaction 42 processing tasks were automated. This allowed RMs to spend more time on providing personalised advisory services to clients. The shift in role led to stronger relationships with clients and increased revenue for both banks. These use cases demonstrate AI’s ability to improve the speed, accuracy, and efficiency of a number of bank processes, whilst at the same time reduce costs and enhance the customer experience (DBS, 2023; White, 2021). Appinventiv’s graph (2024) in Figure 9 below supports the proposition. The digital IT service provider suggested that the market for GenAI in CRM will grow from $27.1 million in 2024 to $119.9 million by 2032, with the biggest opportunity being automated lead generation. Figure 9: Market for GenAI in customer relationship management Source: (Appinventiv, 2024) The proposition aligns to the adopted UTAUT2 framework as performance expectancy, effort expectancy, and social Influence all play key roles in driving technology adoption. According to the theory, individuals are more likely to accept and use a technology if they believe it will help them perform better in their role (performance expectation), as in the case of Deloitte (2024). UTAUT2 also suggests that individuals would more likely adopt a technology if they perceive it to be easy to use (Miah, 2024) and integrate into their existing processes (effort expectancy). This was demonstrated in the DBS bank case (2023). Social 43 influence also plays a role in shaping individuals' attitudes towards technology adoption. The findings from the studies of McKinsey (2023) and Deloitte (2024) suggest that the use of GenAI technology can augment rather than replace the human element in the RM-client relationship (Miah, 2024). This aligns with the theory that the adoption of AI technology in financial institutions can be positively influenced by social factors such as client trust and relationship building. One major implication of this evolution is that RMs will need to develop new skills to effectively leverage AI technology. They will need to be trained in data analysis and interpretation, as well as in communication and interpersonal skills to build and maintain trust with their clients (Miah, 2024). Additionally, RMs may need to adapt their workflow and processes to incorporate AI into their daily activities. Focus will need to shift from transactional work to building relationships and providing valuable advice to clients (Zegullaj et al., 2023). This shift has the potential to greatly transform their roles, placing a stronger focus on the value they deliver to clients rather than simply the quantity of transactions processed. This change is expected to result in heightened client satisfaction and loyalty, ultimately leading to improved retention rates and increased revenue for banks. There are, however, potential drawbacks to consider. One concern is the risk of over-dependency on AI technology, which could result in RMs becoming too detached from their clients (Zegullaj et al., 2023). This could erode trust and hinder the development of meaningful relationships. Another is the concerns about data privacy, regulatory compliance, and potential job displacement (Oracle, 2021). Some may argue that the human touch of RMs cannot be replaced by AI technology, and that customers may prefer interacting with a human rather than a machine. Conversely, AI can complement human capabilities and allow RMs to focus on building deeper connections with customers (Deloitte, 2024). 44 2.7.2.3 Proposition 3 (P3) GenAI adoption in bank lending services presents both opportunities and challenges which must be overcome for successful integration. The core components of the proposition involve acknowledging the potential benefits of GenAI adoption in bank lending, such as improved efficiency, enhanced customer experiences, and personalised services (Kayode, 2024; Mossavar-Rahmani & Zohuri, 2023; Sharma et al., 2023). However, it also requires recognising the challenges associated with integrating AI technology, such as data bias, lack of skills and integration challenges with existing systems (Maple et al., 2023; Oracle, 2021). The proposition holds true because it considers the dual nature of AI adoption that must be addressed, i.e., offering opportunities for innovation and growth, while posing challenges for adoption. The proposition is well supported in empirical research such as a study conducted by Devan et al. (2024) which highlighted GenAI’s benefits in areas like customer service, personalised banking, credit assessment, operational efficiency and predictive analysis. The research indicates that AI-powered chatbots and virtual assistants have revolutionised customer experience by offering instant tailored suggestions. AI algorithms have been effective in reducing risks in credit assessments and loan authorisations, while automation through AI has enhanced operational effectiveness (Bhatore et al., 2020). Predictive analytics has enabled banks to make informed decisions based on data analysis (Emmert-Streib et al., 2020). The study has also shown that despite the advantages of AI in banking, there are concerns about ethics and privacy, emphasising the need for careful consideration of issues like data protection and fairness in algorithms (Domokos & Sajtos, 2024). The authors argue that, whilst AI presents opportunities for innovation and increased efficiency in the banking industry, its responsible use is crucial to mitigate risks and promote fairness. 45 Another case study supporting this proposition, is that of JPMorgan Chase (Weiss, 2017). The bank adopted a GenAI system called COIN (Contract Intelligence) to automate the review of commercial loan agreements. This system helped the bank to process and review loan agreements in seconds, a task that would have taken legal experts 360,000 hours to complete, leading to considerable time savings, enhanced efficiency, lowered operational expenses, and better customer service. Additionally, the system allowed for more customised services by quickly identifying key terms and conditions in loan documents, enabling the bank to offer tailored solutions to their clients. A key challenge is the issue of bias in AI algorithms. Research has shown that AI algorithms can inherit biases present in the data on which they are trained, leading to discriminatory outcomes (Dhake et al., 2024; Kate et al., 2020; Maple et al., 2023; Oracle, 2021). In the context of lending services, this could result in certain groups of people being unfairly denied loans or offered less favourable terms. The findings of Krijger (2023) supports this proposition by exploring the development of responsible AI practices in the banking industry through a case study of De Volksbank. Krijger highlighted the growing concern around the ethical implications of AI use, an example being the risk of bias, resulting in “AI systems that replicate and exacerbate existing disparities by structurally disadvantaging certain groups” (2023, p. 221). The paper examined De Volksbank's framework as a potential model for effectively implementing AI ethics within organisations. It highlighted the importance of organisational design, interdisciplinary expertise, proactive governance, and high-quality processes in addressing AI ethics which are key considerations for this research study. A survey of over 500 senior IT leaders undertaken by Goldman Sachs (BCG, 2023), shown in Figure 10 below, highlighted the concerns and challenges with AI adoption in financial services. 46 Figure 10: C-Suite GenAI adoption concerns and challenges Source: (BCG, 2023) According to the survey, 67% of IT leaders planned to prioritise GenAI within the next 18 months - 33% making it their top priority. However, the concerns of data bias, lack of skills and integration with existing systems that were raised, would first need to be addressed and overcome before they would approve adoption (BCG, 2023). Goldman Sachs listed some of their strategies to overcome these challenges as being: compliance with data governance and ethics, using unbiased data sets, using managed AI services and ensuring necessary training of resources (BCG, 2023) The opportunities and challenges of GenAI adoption in banking services, as highlighted in the studies by Devan et al. (2024), JPMorgan Chase (Weiss, 2017) Krijger (2023) and Goldman Sachs (BCG, 2023), can be grounded in the UTAUT2 framework (Venkatesh, 2022). The opportunities of AI adoption such as customer service, credit assessment, operational efficiency, and predictive analysis can be linked to performance expectancy and effort expectancy. The improved customer engagement and operational efficiency provided by AI-driven chatbots and virtual 47 assistants can increase users' performance expectancy, while the automation of processes can reduce the effort expectancy for employees. The challenges of data bias in AI algorithms, lack of skills and integration with existing systems, can be connected to the constructs of social influence and facilitating conditions. Social influence refers to the impact of social factors on technology adoption, while facilitating conditions refer to the resources and support available to users. In this case, the potential biases in AI algorithms and integration challenges may influence how banks and customers perceive the technology, affecting their behavioural intent to use AI in banking services. Königstorfer and Thalmann (2020) conducted a literature review exploring the applications of AI in commercial banks and the challenges thereof. The review revealed a lack of focus on AI within the commercial banking sector, despite its potential to revolutionise business operations and enhance customer engagement. The authors highlighted the benefits of AI in reducing lending losses (Sharma et al., 2023), streamlining compliance processes (Dhake et al., 2024), and refining customer targeting strategies (Zegullaj et al., 2023) within commercial banks. However, the literature also identified several obstacles to successful AI implementation, such as maximising technological advantages (Mehdiabadi et al., 2020), integrating AI into existing business operations (Mossavar-Rahmani & Zohuri, 2023), fostering user acceptance (Adamek & Solarz, 2023) and addressing data bias concerns (Oracle, 2021). Maximising technological advantages is key as it points to the availability of skilled staff (Mehdiabadi et al., 2020) and the predictive performance of algorithms (Bhatore et al., 2020) as being particularly challenging. The user acceptance challenge speaks to proposition 2 and RMs perception of their evolving role in the era of GenAI, with specific reference to the development of new skills needed to effectively leverage AI technology. The proposition has practical implications for the banking industry, as it highlights the importance of carefully navigating the opportunities and challenges 48 associated with Ge