LegiGPT: Party Politics and Transport Policy with Large Language Model

Political party affiliation significantly shapes transportation policy: The study found that the number and proportion of conservative and progressive bill sponsors were "critical determinants" of legislative outcomes in South Korea's transportation sector, demonstrating that lawmakers' political ideologies directly influence which infrastructure projects and policies get prioritized.

District characteristics matter as much as party politics: Beyond party affiliation, the research identified "district size and electoral population" as equally important factors in determining transportation policy outcomes, suggesting that geographic and demographic considerations can be as influential as partisan ideology in shaping infrastructure decisions.

Both parties contribute through different mechanisms: Rather than one party dominating transportation policymaking, the findings show that "both parties contributed to bipartisan legislation through different forms of engagement, such as initiating or supporting proposals," indicating that understanding how each party participates—not just whether they participate—is crucial for responsible governance.

AI can make political bias in policymaking transparent: By combining large language models with explainable AI techniques, this framework successfully identified hidden patterns in how political ideology influences legislative decisions, offering a replicable tool for citizens and oversight bodies to scrutinize whether policy decisions reflect genuine public needs or partisan interests.

Transparency tools are essential for democratic accountability: The study demonstrates that AI-powered analysis of legislative data can reveal the relationships between political factors and policy outcomes, enabling more informed public debate about whether transportation investments serve all communities fairly or reflect partisan priorities.