Web search engines have evolved drastically over the past two decades, transitioning from simple link providers to direct answer providers. AI technologies, particularly generative large language models, have accelerated this shift by embedding conversational and personalized features directly into search systems. As a result, user expectations and approaches to search have fundamentally changed. Despite this evolution, research continues to rely on search intent frameworks from the early web era. In particular, Broder's influential taxonomy still guides much domain research. Given the transformation of web search, we argue that these frameworks need rethinking. We challenge Broder's taxonomy and propose a new, user-centered framework grounded in contemporary practices. Through an innovative survey combining user reflections with interaction histories, we distinguish three intent categories: knowledge-seeking, guidance-seeking, and output-seeking. This taxonomy applies across search engines and GenAI chatbots, offering a flexible lens for understanding information seeking in different search systems.
ACM CHI Conference on Human Factors in Computing Systems