Engaging the Public Through AI: When is Early Chatbot Adoption effective

• In Event: Infrastructure, Space, and Digital Governance

Abstract

As the next frontier of digitalization (Bankins et al., 2024), artificial intelligence (AI) technologies are beginning to take root within public organizations (Neumann et al., 2024; Wirtz et al., 2019). AI-enabled chatbots have shown great potential for improving public participation and stakeholder engagement through 1) increased efficiency of information provision and citizen-state communication (Larsen & Følstad, 2024; Pencheva et al., 2018; van Noordt & Misuraca, 2022), and 2) enhanced interaction experience that allows for richer and more expressive expression of citizen needs (Androutsopoulou et al., 2019; Chaves & and Gerosa, 2021; Gagliardi et al., 2017). While researchers are optimistic about the role of AI chatbots on improving public participation and government responsiveness, transparency and accountability, their impacts on participation outcomes remains empirically underexplored. To address this gap, this paper asks the question: under which conditions does AI chatbot implementation enhance public participation in local government? Drawing on socio-technical theory, this paper hypothesizes that the effect of AI chatbot is shaped by the interplay between several organizational factors, including organizational structure and organizational innovativeness, and the economic and political environment in which it operates. Socio-technical theory underscores the contextual nature of digitalization and holds that the effectiveness of a technology is interdependent on the organizational and institutional contexts in which it is implemented (Mumford, 2006; Tornatzky & Fleischer, 1990). To test my hypotheses, I use merged data from 1) a national survey on public managers of six departments (community development, public works, human resources, finance, police, parks and recreation) from local governments across 650 U.S. cities with a population less than 300,000 the conducted by the Center of Science, Technology and Environmental Policy Studies (CSTEPS) at ASU in 2024, and 2) American Community Survey from U.S. Census Bureau, to estimate the relationships between organizational characteristics, social context and the effectiveness of chatbot adoption. This paper employs a regression model to investigate the participatory outcomes of local government, as measured by the frequency of engagement with different groups of stakeholders, and the type of inputs provided. Moving beyond accounts that focus solely on technological capability or institutional pressures, findings highlight that the impact of early AI adoption is contingent upon the alignment between internal organizational structures and external contextual conditions. They also offer a more nuanced understanding of when and why automated interactions with chatbots succeed in engaging the public.

 

Author

• Jinghuan Ma, Arizona State University
  Presenting Author