California Uses AI to Process Public Policy Feedback
- •California is using its Engaged California platform to collect public feedback on AI's impact on employment.
- •ODI utilizes large language models to categorize public input and generate actionable policy insights from unstructured data.
- •The initiative includes a second phase of live forums this summer and a final report scheduled for September.
California is utilizing its digital conversation platform, Engaged California, to gather and analyze public input on how artificial intelligence is affecting the workforce and economy. The initiative, managed by the Office of Data and Innovation (ODI), marks the third project for the platform since its first pilot in February 2025. Previous pilots addressed aid for Los Angeles wildfire victims and gathered state employee feedback on government efficiency. The current AI-focused engagement officially launched on May 7, 2026, with an emphasis on timely execution to maximize public participation.
The methodology employs large language models (LLMs) in two distinct roles: querying comments in plain language to identify emerging themes and synthesizing feedback into a structured final report. An internal dashboard enables ODI officials to interact with raw, unstructured data to extract actionable policy recommendations. According to ODI Director Jeffery Marino, the platform serves as a digital town hall, supplementing rather than replacing existing mandated public comment processes. The project intends to compile findings in a public report scheduled for release in September 2026.
In addition to the initial digital feedback phase, a second stage is planned for later this summer, featuring live forums where a representative group of California workers will discuss potential policy solutions. To ensure transparency, the project makes its raw data, code, and methodology available via open source. ODI is also collaborating with other agencies, such as the California Environmental Protection Agency, to develop similar tools for addressing backlogs of public comments. Lead product manager Jarrett Krumrei suggests that other governments adopting this model should collaborate with in-house IT teams and conduct user testing within secure, isolated sandbox environments (controlled testing areas for software).