(iPoster) Can Governments Govern by Robots? Citizens’ Cooperation in Different Scenarios

• In Event: Poster Sessions: Information Technology and Politics & Political Networks
  In Poster Session: Poster Session: Information Technology and Politics

Abstract

Background
The rapid diffusion of artificial intelligence has enabled governments to deploy AI-powered service robots as front-line public service providers. These systems are increasingly framed as tools for improving administrative efficiency, service equity, digital governance and promoting democracy. In multi-government settings, citizens encounter government robots across diverse administrative contexts. Yet despite their growing presence, systematic evidence remains limited on whether citizens are willing to cooperate with robot civil servants and how robot characteristics shape such cooperation across governance scenarios.
Existing research on service robots focuses primarily on commercial settings or text-based chatbots, with limited attention to embodied robots in the public sector. Moreover, citizen–government interactions are often treated as context-neutral, overlooking how variations in urgency, conflict, and normative expectations shape cooperation.
This paper addresses these gaps by examining whether citizens’ cooperation with government robots varies across governance scenarios and how such cooperation is shaped by robot’s external design features and AI’s internal social attributes in different governance contexts.
Conceptual Framework
This study conceptualizes citizens’ cooperation preferences as individuals’ willingness to engage with and rely on government robots in interactions with public authorities. Cooperation is understood as a relational outcome shaped by expectations of legitimacy, responsiveness, and procedural fairness in digital governance.
Multi-government scenarios denote distinct governance contexts varying in urgency, conflict, and normative expectations, capturing heterogeneity in citizen–state interactions from routine services to high-stakes encounters.
The study distinguishes between external design features—the visible and physical attributes of government robots that shape symbolic perceptions of authority—and internal social attributes, which denote AI-enabled interactional capacities such as responsibility, communication competence, and emotional responsiveness. Together, these dimensions provide a framework for examining how cooperation with government robots is formed across governance contexts.
Research Design
This study employs a multi-modal experiment built around a discrete choice experiment (DCE). To reduce respondents’ cognitive burden, the experiment integrates text-to-image generation (Stable Diffusion) and large language model–based scenario construction (an AI agent based on ChatGPT4.0). Generative AI techniques are used to transform abstract governance scenarios and robot attributes into coordinated visual and dialog-based stimuli, enabling participants to make choices in settings that more closely resemble real-world interactions in digital governance.
The empirical analysis draws on data collected from 5000 participants across China and covers multiple governance contexts, including information provision, personal service delivery, emergency response, and rights-protection scenarios.
Analytically, a two-stage approach is adopted. First, logistic regression is used to estimate interpretable main effects of robot characteristics on citizens’ cooperation preferences. Second, a multilayer perceptron model is applied to capture nonlinear and interactive patterns in preferences, serving as a robustness check for the findings derived from conventional econometric analysis.
Results and Discussions
Results indicate that citizens prefer government robots with cooler color tones, friendly body styles, high responsibility, and stronger emotional intelligence, while displaying resistance toward proactive service recommendations and explicit display screens. Cooperation intentions are significantly higher in non-urgent and positively framed scenarios such as policy consultation and routine administrative services.
In contrast, in urgent or conflict-oriented contexts—such as complaints and accountability—citizens continue to favor human officials, reflecting persistent expectations of discretion and emotional responsiveness in high-stakes governance interactions.
Contributions
By integrating citizen preference theory with research on AI-enabled governance, this study contributes to political science and public administration at the theoretical, methodological, and empirical levels. Theoretically, it conceptualizes government robots as embodied governance actors at the front line of citizen–state interactions, extending digital governance research beyond algorithmic tools. Methodologically, it combines machine learning techniques with conventional econometric models to analyze context-dependent cooperation preferences. Empirically, using a large-scale discrete choice experiment, the study provides systematic evidence on how citizens’ cooperation with government robots varies across governance scenarios.

Authors

• Unregistered Participant

• Unregistered Participant