PROBLEMÁTICA DO ATENDIMENTO DE SUPORTE AO CLIENTE POR CHATBOTS: UM ESTUDO EMPÍRICO DO GPT-4

The present article is an empirical study of Banco Inter’s database and a case study of a software startup called Phibo, focused on its problem-solving capacity. It evaluates its functionalities in comparison with the capabilities of existing chatbots and service robots in the customer service scenario. The methodology employed in this study was a mixed quantitative and qualitative approach, conducted through case studies of the companies Phibo and Banco Inter. The study showed significantly superior performance in various key metrics compared to its predecessors and other Artificial Intelligence (AI) systems commonly used for customer support. Furthermore, the research identified that although GPT-4 has achieved impressive performance, issues such as AI ethics, transparency in decision-making processes, and the continuous need for training and updating the models remain critical challenges to be addressed in the future.