KOTI - Korea Transport institute

Theme: Spatial Analysis and Applications Using Generative AI

The KOTI Mobility Colloquium series is designed to anticipate the future of mobility services driven by transportation technology innovation and to identify key research and development (R&D) topics necessary to realize that future.

This session explored the potential of automating spatial information interpretation—such as complex urban environments and building regulations—by leveraging large language models (LLMs). The discussion also covered ways to enhance search accuracy, reasoning precision, and response reliability through improvements in vector databases and question-answering (QA) structures.

The outcomes demonstrated that generative AI can significantly reduce the time and improve the accuracy of legal and regulatory interpretation, proving its practical value. Furthermore, its adaptability in spatial analysis was confirmed, highlighting the expanding potential for AI-driven applications in future urban and architectural research.

Presentations:

- AI-based Urban Spatial Information Search and Optimization of Office Space Utilization / Prof. Han Jeong-Hoon, UNSW Sydney

- A Legal Interpretation Support System for Building Codes Using Large Language Models / Dr. Cho Sang-Kyu, Architecture & Urban Research Institute (AURI)