Using Large Language Models in Public Transit Systems, San Antonio as a case study

TL;DR

This study investigates deploying large language models in San Antonio's public transit using GTFS data to enhance route planning, real-time communication, and resource allocation. It adopts a two-task framework—GTFS understanding and GTFS retrieval—evaluating GPT-3.5-Turbo and GPT-4o on a curated, city-specific QA dataset with a GTFS Retrieval benchmark, while addressing context-length limitations by trimming data to three routes. Results show GPT-4o generally outperforms GPT-3.5-Turbo across both tasks, delivering strong performance on structure and mapping categories but varying success on more abstract mappings, and notably improving retrieval accuracy. The findings illustrate substantial potential for LLMs to improve transit services, while highlighting gaps, ethical considerations, and the need for careful, ongoing evaluation before real-world deployment. The San Antonio VIA case provides a scalable blueprint for other cities exploring LLM-enabled transit solutions.

Abstract

The integration of large language models into public transit systems represents a significant advancement in urban transportation management and passenger experience. This study examines the impact of LLMs within San Antonio's public transit system, leveraging their capabilities in natural language processing, data analysis, and real time communication. By utilizing GTFS and other public transportation information, the research highlights the transformative potential of LLMs in enhancing route planning, reducing wait times, and providing personalized travel assistance. Our case study is the city of San Antonio as part of a project aiming to demonstrate how LLMs can optimize resource allocation, improve passenger satisfaction, and support decision making processes in transit management. We evaluated LLM responses to questions related to both information retrieval and also understanding. Ultimately, we believe that the adoption of LLMs in public transit systems can lead to more efficient, responsive, and user-friendly transportation networks, providing a model for other cities to follow.

Figures (2)

• Figure 1: Summary of performance by question category for GPT-3.5-Turbo and GPT-4o on GTFS Understanding.
• Figure 2: Summary of performance by question type for GPT 3.5-turbo and GPT-4o on GTFS Retrieval Benchmark