A Theoretical Framework for AI-Assisted Civic Issue Reporting and Validation in Smart Cities

Urban administrations face persistent challenges in reporting, prioritizing, and validating civic infrastructure issues due to manual workflows, inconsistent citizen inputs, and limited resolution verification mechanisms. Existing digital grievance systems remain largely text-centric and require substantial human intervention. This paper presents a multimodal AI–assisted, crowdsourced civic issue management framework that leverages Multimodal Large Language Models (MLLMs) for image-based issue understanding and workflow support. Citizens submit geotagged images through a mobile-friendly interface, enabling automated generation of standardized descriptions, semantic categorization, and severity-aware prioritization to assist municipal decision-making. A key contribution is an AI-assisted before–after image validation mechanism that formulates issue resolution as a semantic change detection task, allowing meaningful repairs to be distinguished from irrelevant visual variations. Rather than proposing new model architectures, the framework demonstrates system-level integration of existing multimodal AI capabilities to improve transparency, consistency, and accountability in smart city governance while maintaining human oversight.