Project Summary
California invests in public engagement on transportation policy, yet translating that input into policy that reflects public values remains hard. Existing mechanisms each hit structural limits: public comments lack synthesis, town halls do not scale, and surveys miss the deliberative depth needed for complex tradeoffs. The research team proposes to address this gap using a custom AI-assisted democratic deliberation tool developed by the team’s lab. The platform collects private opinions from large groups, synthesizes them into candidate consensus statements using large language models, and iterates through participant feedback to produce validated group positions that preserve minority viewpoints — combining the scale of surveys with the deliberative depth of citizen assemblies. In pilot studies across 10 policies (N=300), policies revised through this process showed an average improvement in public support of 1.5 points on a 10-point scale over original versions. This project scales this approach to California’s transportation policy context: N=4,000 California residents deliberate in groups of 100 across four policy domains (Road User Charge, EV rebates, feebates, bike infrastructure), and N=1,000 independent raters evaluate original versus deliberation-revised policies. Results will produce actionable policy guidance for Caltrans, a replicable methodology for human-centric policy design and AI-assisted deliberation as a governance tool.
