Abstract:
The study aims to investigate the capabilities of GPT-4, a cutting-edge large language model developed by OpenAI, in the context of transportation planning in Sri Lanka. This research delves into three main sectors: transport planning knowledge, concepts, and strategies; technical-oriented data, methodologies, and tools; and stakeholders' opinions. The study examines GPT-4's ability to articulate transport planning concepts and methods, providing creative policies adapted to Sri Lanka's specific challenges using a thorough approach. The model's capacity to provide methods for data analysis using Geographic Information Systems (GIS) applications is also investigated. The findings reveal that GPT-4 demonstrates outstanding capabilities in proposing practical transportation strategies and policies, akin to those presented by experienced planners. Additionally, it shows a high level of proficiency in technical data processing, expediting the decision-making process. Despite acknowledging its drawbacks, the GPT-4 needs to be improved using Sri Lanka-specific domain information in order to improve contextual knowledge and provide accurate results. Understanding the influence of training data and sourcing relevant information is crucial for specialized applications like transportation planning. Planners must be aware of GPT-4's limitations, particularly in detecting hallucinations that simulate coherent responses. Human planners should maintain their role as critical decision-makers, leveraging AI tools like GPT-4 while continuing to invest in competencies, cross-domain knowledge, and data analysis proficiency. Striking this balance ensures effective collaboration in the transportation planning process.
