This paper is grounded in the context of modernizing national governance and explores how artificial intelligence leverages its technical potential to drive evidence-based governance in government. Based on the evolutionary trajectory from evidence-based decision-making to evidence-based governance，the study constructs an analytical framework covering the entire governance process agenda setting，policy formulation，implementation and feedback. It systematically elaborates on the enabling role of artificial intelligence in data aggregation，evidence generation，and the construction and application of dynamic evidence repositories，while revealing its mechanisms for reconstructing the logic of evidence across three dimensions：technology，knowledge，and objectives. Furthermore，from a holistic process perspective，the study examines the pathways through which artificial intelligence facilitates the formation of a data-intelligence-driven，proactive governance paradigm. While expanding the theoretical and operational boundaries of evidence-based governance in government，artificial intelligence also introduces challenges such as algorithmic bias，data security，and public accountability. Future efforts should focus on building a more forward-looking，responsive，and resilient smart governance system to promote the realization of public values，balance instrumental rationality and value rationality，and continuously advance the modernization of government governance capacity.