Facilitating Urban Participation Project with Generative AI to Support Citizen Engagement and Interaction

Abstract

This paper investigates how AI-based chatbots can enhance citizens’ engagement and interaction on urban participation platforms. Using a design science research approach, we identified twelve issues, formulated eleven meta-requirements, and derived five design principles. These were instantiated with a web prototype designed in Flutter, utilizing a large language model, including interaction and expressing guidelines. Our evaluation revealed increased engagement, lower participation barriers, and improved citizen contributions compared to non-AI-based participation. However, the evaluation also led to the addition of two new ones, highlighting document access and interactive urban maps. Together, they specify information presentation and interaction with participants. Despite promising findings, challenges persist regarding the perception and explainability of large language models. Our findings provide a practical blueprint for future AI-enabled citizen participation in urban planning, suggesting directions for further research on the retrieval-augmented generation architecture, which can incorporate additional domain knowledge and behavioral guidelines.