The Taiwan Banker

As artificial intelligence becomes deeply embedded in banking, one veteran employee of Commonwealth Bank of Australia found herself in an ironic situation: she had unknowingly trained the very system that would replace her. Kathryn, a 63-year-old teller with 25 years of service, was laid off after helping to design and test chatbot responses for the bank’s AI customer service project, Bumblebee AI. As she put it, “Inadvertently, I was training a chatbot that took my job.”After the incident drew public attention, the bank was forced to admit its misjudgement. Instead of reducing workload, the launch of the AI system led to a surge in customer service calls; management eventually reversed the layoffs and issued a public apology. This episode highlights a global reality in the financial sector: rather than improving service quality, AI introduced under the banner of efficiency, without clear human-machine role division or ethical design, can erode customer trust and professionalism. True intelligent customer service does not mean replacing humans with machines. Rather, AI should be used as an engine for both service and talent upgrades – for instance, handling information-processing tasks so that humans can focus on relationship-building and trust. Only through complementary coexistence between humans and machines can banks strike a new balance between efficiency and empathy.In the past, customer service chatbots were mainly used to reduce workload, but they have evolved from simple automated responses to proactive assistance. With advances in generative AI and large language models (LLMs), such systems are now becoming the core knowledge base and operational hub of banking services. Standard Chartered Bank, for instance, built an intelligent customer service system early on based on KAI, a platform developed by U.S. company Kasisto which instantly handles common queries such as balance checks, card loss reports, exchange inquiries, and credit card application status. This system greatly improved efficiency, yet frontline service staff did not disappear. They remained an essential second line of human support behind the system. The essence of smart customer service lies in human-AI collaboration. When customers become anxious, file major complaints, or face cross department issues, the AI automatically triggers human intervention. At that moment, the agent enters a collaborative interface where they see suggested replies, customer histories, and system-generated indicators. They can choose to accept, modify, or override the suggestions. Here, the human value lies not in speed, but in sensitivity and empathy – qualities AI still lacks. They can interpret the emotional undertones behind a customer’s words and respond with reassurance or compensation when needed. This evolution in collaboration has also reshaped how frontline employees view their roles. No longer mere call operators but directors and advisors, even as AI can anticipate customer needs and search for answers autonomously, they must decide whether, when, and how to step in. Sometimes, a single sincere word of comfort is more effective than any automated reply; sometimes, the best service is simply to listen. AI delivers speed, but humans deliver warmth. Process redesign is essential for this coexistence model to truly create value. Banks must clearly define which tasks are for AI and which require human oversight, establishing transparent division labor and monitoring points. A workflow structure such as AI pre-views, human verification, and exception-handling matrices should become standard practice. Yet no matter how sophisticated AI systems become, they still depend on improvements in human capability and organizational culture. Talent transformation marks another challenge for banks in the AI area. Competition is no longer about which bank has the most advanced technology, but who nurtures their talent faster and more effectively. Frontline staff must master AI operation, data interpretation, and emotional regulation. Managers, in turn, need to design human-AI co-governance processes to ensure that AI serves as a decision-making assistant rather than a source of risk. Leading international banks are already addressing this challenge through various forms of employee reskilling. For instance, JPMorgan Chase has made AI literacy a core part of its employee training. New hires are required to learn prompt engineering, gaining the ability to collaborate with generative models, and shifting their roles from executors to AI orchestrators. Standard Chartered has adopted a skills-based organizational structure, making AI, digitalization, innovation, and leadership core competencies. The bank also holds an annual “Al and Data Week” to promote continuous learning and internal mobility. DBS Bank began cultivating digital literacy a decade ago. Its DBS-GPT platform now assists employees in data retrieval, translation, and decision-making. As automation deepens, traditional roles have evolved into entirely new ones, for example customer service representatives becoming “digital journey designers,” and loan officers transforming into “digital credit process optimization consultants.: While AI can deliver efficiency and rule-based rigor, only humans can perceive emotions and bear ethical responsibility. In this wave of transformation, banks should let machines handle information and rule execution, while the people focus on building relationships, understanding anxiety, interpreting value, and making judgements. The key to human-AI coexistence lies in this balance: let machines drive efficiency, and humans safeguard meaning. The author is a Professor in the Department of Information Management and Finance at National Yang Ming Chiao Tung University and an Independent Director at Hua Nan Financial Holdings.