Abstract: [Purpose/Significance] This study explores the characteristics manifestations of users' search behavior, cognitive pathways, and cognitive transitions during information-seeking processes in a generative artificial intelligence environment, and compares them with those of traditional search engine users. The aim is to provide references for the research on search as learning features and the improvement of search tools in the context of generative artificial intelligence. [Design/Methodology] Through a user experiment, participants were divided into a traditional search engine group and a generative artificial intelligence group. Data were collected using screen recording and think-aloud method, followed by data analysis based on non-parametric tests analysis method. [Findings/Conclusion] Significant differences are observed between the generative artificial intelligence group and the traditional search engine group in terms of search behavior, cognitive pathways, and cognitive transformation. Generative artificial intelligence deepens users' search depth and increases search width, but does not promote more complex learning for users. [Originality/Value] Incorporating generative artificial intelligence search tools into the research scope of "search as learning", this study deeply analyzes the differences in users' search behavior and cognitive path when using generative artificial intelligence search tools versus traditional search tools. It expands the research context and field of "search as learning", helping designers of generative artificial intelligence to assist users in better using natural language for information query and learning.

Keywords: Information seek, Cognitive path, Generative artificial intelligence, Search as learning