Embodiment of AI Governance: How Government Digital Humans Reshape Public Acceptance of Digital Public Services

In the grand narrative of digital transformation, generative AI is penetrating the public service sector at an unprecedented speed, and government digital humans are the embodied manifestation of this integration. This is not just an upgrade of technological tools, but a profound shift in the paradigm of urban governance from "fixed programs" to "intelligent interactive entities." From the tax digital human "Yue Yue" introduced in Guangzhou in China to smart service robots in other cities, these digital humans utilize generative AI, cloud computing, and big data to provide round-the-clock, highly personalized administrative approval, intelligent Q&A, and business processing services in the form of virtual avatars or intelligent robots. Their core value lies in breaking the rigidity of traditional fixed systems, transforming complex government processes into intuitive, convenient conversational experiences through anthropomorphic interaction.

However, a leap in technological capability does not automatically translate into public willingness to adopt. As academic research indicates, the key to successfully embedding generative AI in public services is no longer just the advancement of the technology itself, but the public's psychological expectations and trust in this 'digital agent'. This study aims to deeply couple Expectation Confirmation Theory (ECT) with the Technology Acceptance Model (TAM) to move beyond traditional "supply-side" efficiency arguments toward understanding the acceptance logic on the "demand side."

We find that public willingness to accept government digital humans is not driven by a single factor. It is jointly influenced by a series of complex mediating variables. First, 'expectation confirmation'—that is, the user's expectation of the service quality and interaction experience this digital human can provide—is the starting point of the acceptance process. Second, this expectation is mediated through a series of perceived dimensions. Perceived 'intelligence', 'information quality', and 'anthropomorphism' are key in shaping the digital human's image and functional reliability. When a digital human is perceived as intelligent enough, providing accurate information, and exhibiting enough 'humanity' in interaction, its 'attractiveness' and 'usefulness' will significantly increase.

The deeper driving force lies in 'AI trust'.A deeper driving force lies in 'AI trust'. In the age of data and algorithm-driven systems, public trust in AI is the final line of defense determining whether they are willing to entrust sensitive matters to digital humans. Trust is built on transparency, reliability, and a clear definition of privacy security. Therefore, for urban governance departments, optimizing digital human services should not merely be about improving the parameters of its AI model, but should focus on enhancing its 'perceived usefulness' and 'trustworthiness'. This requires governors to shift from the level of technical implementation to the systematic design of the 'digital experience', ensuring that every human-machine interaction finds the optimal balance between efficiency, accuracy, and a sense of trust.

This shift in research paradigm marks an upgrade in our understanding of future urban infrastructure. The future city will no longer just be a pile of concrete and steel, but a service network driven by AI and equipped with embodied intelligence. Urban systems must learn how to design 'digital public spaces', making technology not just a backend optimization engine, but a trustworthy, perceptive service partner facing citizens. This demands that when advancing the construction of the 'urban operating system', 'user expectations' and 'technological ethics' must be placed on an equally important strategic level, ensuring that technological progress ultimately serves to enhance citizen well-being and digital sovereignty.