Artificial Intelligence has developed steadily over recent decades. Because of the swift adoption across various applications, particularly generative AI, AI techniques have been adopted in many industries. One prominent application is the AI-powered chatbot which is commonly used to improve operational efficiency, enhance customer service, personalize and tailor responses, improve cost effectiveness, and so on. This study aims to evaluate AI chatbots implemented by top banks in the United States. Systematic and comprehensive analyses are conducted on these AI chatbots using the same pre-designed tasks. The chatbots are evaluated based on their precision, accuracy, and failure. Advantages and disadvantages of these AI chatbots are identified. Suggestions for the future improvements of these chatbots are provided. Overall findings suggest that the integration of AI into bank chatbots should not be seen as a binary shift from human to machine, but rather as a complementary tool that enhances, yet does not replace human capital.
