How Amazon’s AI Assistant Rufus Affects User-Generated Q&A

User-generated Q&A functions on e-commerce platforms allow potential consumers to ask specific questions before making a purchase to address their individual information needs. Compared to traditional static online reviews, Q&As not only reduce pre-purchase uncertainty but also enhance product-consumer fit (Banerjee et al., 2021). However, the recent introduction of Amazon’s generative AI assistant Rufus may further alter how consumers seek and share information. By providing real-time, structured answers directly to the customers, Rufus significantly reduces the information search cost and potentially diminishes consumers' willingness to ask new questions or engage in existing Q&A interactions. Recent studies have begun exploring the impact of large language models (LLMs) like ChatGPT on knowledge-sharing communities, suggesting that ChatGPT’s launch led to a measurable decline in user question-asking behavior on Stack Overflow (Burtch et al., 2023). However, these insights have primarily focused on technical domains and community-oriented platforms, limiting their generalizability. While online communities are typically designed to foster social interactions, enhance willingness to share knowledge, and build collective intelligence (Chiu et al., 2006), the primary purpose of Q&A functions in e-commerce settings is to facilitate sales conversions by reducing purchase uncertainty. Therefore, the influence of generative AI on user knowledge-sharing behavior may vary substantially between community-oriented platforms and e-commerce platforms. To extend this line of research to the e-commerce context, we focus on Amazon’s Rufus and further examine potential heterogeneity across different product types, specifically search and experience goods. This distinction matters because search goods can be evaluated through objective information such as specifications and parameters, while experience goods rely on actual sensory experiences for evaluation (Nelson, 1970). To sum up, this study aims to answer the following research questions: (i) how does the introduction of a generative AI assistant (Rufus) influence user interactions in Q&As on e-commerce platforms? (ii) how does this influence differ across product types, such as search and experience goods? This study employs a Difference-in-Differences (DID) approach to identify its impact on Q&As in e-commerce platforms. Specifically, it explores variations in question volume, question quality, as well as answer volume, answer quality and answer latency. The findings will offer insights into how AI assistants such as Rufus affect knowledge-sharing dynamics in e-commerce platforms.