Abstract: The intelligent transformation of public services is a significant manifestation of the new round of technological revolution. Existing literature mostly focuses on the level of organizational change, yet it fails to deeply explain the underlying logic of artificial intelligence (AI) driven public service transformation when describing the details. Drawing on the theoretical inspiration of the Actor-Network Theory (ANT) and based on the perspective of human-machine relationships in public service transformation, this study constructs an analytical framework for AI-driven public service transformation and examines its impact on service efficiency, service scenarios, and public value. The findings reveal that AI accelerates service innovation through comprehensive improvements in decision-making efficiency, collaborative efficiency, allocation efficiency, and communication efficiency; it also promotes the upgrading of service space by restructuring the elements, representations, atmosphere, and value of scenarios, thereby shaping an entirely new form of public services. However, while creating public value, AI simultaneously exacerbates the risk of value failure in service delivery due to its inherent vulnerabilities in trust, accountability, ethics, and emotional capacity. In response, it is necessary to build a human-machine symbiosis paradigm in interaction and reaffirm and establish the dominant position of humans in AI-empowered public services. Deep human-machine collaboration can thus chart a new path for the modernization of public service development.

Key words: artificial intelligence, public services, efficiency revolution, scenario reconstruction, public value failure