Generative AI and Automated Customer Service in E-Commerce

Abstract

Generative Artificial Intelligence (GenAI) is rapidly transforming customer service in electronic commerce by changing how businesses communicate with customers, process service requests, resolve complaints, and provide personalized assistance. Traditional e-commerce customer service has relied heavily on human agents, rule-based chatbots, frequently asked questions, and scripted responses. Generative AI introduces a more flexible model in which large language models can understand natural-language questions, generate context-sensitive responses, summarize customer interactions, retrieve information from business databases, and increasingly support or execute service-related tasks. This transformation has important implications for service quality, operational efficiency, customer satisfaction, personalization, employee productivity, and the economics of digital commerce.

The present paper examines the role of generative AI in automated customer service in e-commerce. It analyses major applications including AI chatbots, conversational commerce, automated complaint handling, order-status assistance, product support, return and refund management, sentiment analysis, agent-assistance systems, multilingual service, and emerging AI agents. The paper argues that GenAI can reduce response times, extend service availability, lower repetitive workloads, improve consistency, and enable businesses to scale customer support across large digital markets. Recent evidence is particularly relevant: an OECD survey of more than 5,000 SMEs found that customer service was one of the significant business-support uses of generative AI, with 37.3% of GenAI-using SMEs reporting use in customer service.

Nevertheless, automation creates significant risks. Generative AI can produce inaccurate or fabricated responses, mishandle sensitive information, misunderstand complex complaints, reproduce bias, and create uncertainty concerning accountability. Excessive automation may also weaken human empathy and reduce customer trust, particularly in emotionally sensitive or high-complexity situations. Recent field evidence from e-commerce after-sales operations suggests that GenAI assistance can improve service speed and some measures of customer-rated quality, while effects can differ across worker groups and task contexts.

The paper concludes that the most sustainable model is not complete replacement of human customer-service employees but human-AI collaboration, in which generative AI handles routine, information-intensive, and repetitive interactions while human agents manage complex, sensitive, exceptional, and high-value cases. Responsible implementation therefore requires accuracy controls, human escalation, privacy protection, transparency, continuous monitoring, employee training, and clear accountability.