Dan Milz, Jennifer Kagan, and I wrote an opinion piece for Honolulu Civil Beat called “With Hawaii Civic Engagement At A Crossroads, Let’s Use AI For Good”.

We wrote it because of a paradox in Hawaiʻi’s civic life. The state has led on transparency and public participation, and yet the sheer volume of testimony now outstrips anyone’s ability to absorb it. One 2023 Commission on Water Resource Management meeting ran more than 12 hours, 9.5 of them public testimony. Much of what government does is getting buried in an avalanche of transparency.

Our argument is that generative AI can help here, not by replacing deliberation but by synthesizing and categorizing unstructured input from meetings, surveys, and public comment so that decision-makers can engage with what people actually said. We called on Hawaiʻi’s leaders to convene a statewide conversation among agency officials, elected representatives, researchers, and citizen advocates about how to do this responsibly and what policy changes it would take.

This line of thinking continues in our research on analyzing participatory planning meetings with large language models.