Towards Intelligent Shopping Assistant: Can LLM Chatbot Empower Consumer Decision Making?

Abstract

To explore the potential of Generative AI to enhance consumer decision making in E-commerce, we build an online shopping platform where consumers can conduct product searches and make purchase decisions, and design a chatbot assistant, powered by Large Language Models (LLM), that can help them shop on the platform. We conduct incentive-aligned experiments to evaluate the impacts of the AI chatbot on consumer shopping behaviors. Contrary to our expectation that the chatbot assistant can help consumers process information more efficiently and thus expedite the shopping process, we find that it actually increases consumers' shopping time. We further explore the underlying mechanism and find that the chatbot can encourage consumers to search more intensively (i.e., within-product inspection) and search with deeper concentration, while also marginally prompting them to search more extensively (i.e., cross-product search). Furthermore, the chatbot can help consumers reduce information overload, improve their shopping experience, and increase their purchase likelihood.

Keywords: Generative AI, LLM, chatbot, consumer search, information overload