Generative AI in Transportation Analytics: Foundations, Deep Generative Models, and Practical Use Cases

Generative artificial intelligence (Gen-AI) is rapidly reshaping transportation analytics by enabling data synthesis, uncertainty modeling, and decision-oriented reasoning under sparse and complex conditions. This study presents a systematic, use-case-driven review of 142 empirical transportation studies published between 2018 and August 2025, synthesizing how Gen-AI has been adapted, evaluated, and integrated into real transportation workflows. This study distinguishes two dominant model families, deep generative models and foundation models, and show that they serve fundamentally different yet increasingly complementary roles: the former addressing data scarcity and distributional learning, and the latter enabling reasoning, coordination, and decision support. Through bibliometric analysis, topic modeling, and application mapping, this study identifies major adoption trends, persistent limitations, and research gaps related to generalization, validation, scalability, and trust. The findings reveal a field-level shift from prediction-centric modeling toward generative, decision-aware transportation intelligence and outline a roadmap for responsible, domain-grounded deployment of Gen-AI in safety-critical transportation systems.