A study on online customer service enhanced by artificial intelligence

Abstract

In today's digital age, online customer service is playing an increasingly vital role in corporate service systems. This study examines the business-to-business (B2B) scenarios of banking and offers an in-depth analysis of four online customer service modes: the AI only mode, the AI-human-backup mode, the AI-human-partner mode, and the human only mode. The primary aim is to explore the impact of these service modes on customer satisfaction and human agents’ workload. This study employs an experimental research method based on real online customer service scenarios in banking. A random sample of customers was selected, and questionnaire data were collected and analyzed. The results indicate significant differences between the AI only mode and the three human-involved service modes (i.e., the AI-human-backup mode, the AI-human-partner mode, and the human only mode) in dimensions related to customer satisfaction. The three human-involved modes performed better in terms of perceived competence, perceived warmth, and perceived accountability, leading to higher overall customer satisfaction compared to the AI only mode. However, no significant differences were observed among the three human-involved service modes in terms of customer satisfaction and the underlying mechanisms influencing satisfaction. Further analysis of the workload of human customer service agents within the three human-involved service modes revealed that, under the same level of customer satisfaction, the AI-human-backup mode was more efficient in reducing the workload of human agents compared to the human only service mode and the AI-human-partner mode. This efficiency suggests that the AI-human-backup mode can effectively reduce labor costs for enterprises. This study fills a gap in existing research by providing a detailed exploration of hybrid service modes and presents new perspectives on the development of online customer service systems. It contributes to the theoretical framework in this field and lays a foundation for future research. Additionally, the research analyzes the impacts of service modes on perceived competence, perceived warmth, perceived accountability, and satisfaction, and investigates how these factors mediate customer satisfaction. The study also explores the impact of different service modes on the workload of customer service representatives, offering a comprehensive analytical framework. Furthermore, this study uses large language model(LLM) to analyze chat log with a larger sample, which finds that the two hybrid modes (i.e., the AI-human-backup mode and the AI-human-partner mode) outperformed the AI only and human only modes, and could reduce human agents’ workload compared to human only mode. It also provides a preliminary exploration of AIGC (AI-generated content) applications in the banking sector, demonstrating its potential to enhance intelligent customer service. This offers practical insights for banks and other industries seeking to leverage AIGC technology to optimize their customer service. Keywords: customer service modes, customer satisfaction, competence, warmth, accountability, human agents’ workload, LLM, AIGC