Abstract: [Purpose/Significance] The study aims to enrich the understanding of human–AI interaction behaviors in the context of Generative Artificial Intelligence(GenAI), and to provide suggestions for cultivating prompt literacy among humanities and social science researchers. [Design/Methodology] Based on metacognitive theory, this study conducts a longitudinal tracking study using the diary research method . The interview data from 25 humanities and social science researchers are coded and analyzed by drawing on the methodology of Gioia and subsequently a model is constructed. [Findings/Conclusion] The evolution of prompting behavior among humanities and social science researchers shows distinct phased characteristics, including exploration and adaptation, strategy construction and regulation, transfer and innovation. The study also identifies three types of prompting behaviors: exploratory, structured, and high-level prompt design, as well as the corresponding three stages of adaptive information interactions, strategic information interactions, and collaborative co-creation interactions. Furthermore, metacognitive knowledge, experience, and regulation have an impact on the cultivation of prompt literacy at different stages. Based on this, this study constructs an integrated model of "Metacognitive Activities–Prompting Behavior–Prompt Literacy". [Originality/Value] The study extends the application boundaries of metacognitive theory in the context of human-AI interaction and provides both theoretical foundations and practical implications for cultivating prompt literacy among humanities and social sciences researchers.

Keywords: Prompt literacy, Prompting behavior, Metacognitive Theory, Humanities and social science researchers, Human-AI interaction