Reflections on AI, Democracy, and the Work Ahead

Thanks again to Metagov for the Seminar invitation—it was a pleasure to share the work we’ve done to track and understand Civic AI.

We’re increasingly focused on how AI is showing up across civic engagement, advocacy, govtech, and beyond, for better or worse.

In this talk, I shared a slice of that work, including our recent deep dive into 30 digital participation platforms using AI to support everything from translation and summarization to new formats for storytelling and feedback (together with People Powered).

I also highlighted some standout examples like Thinkscape and UrbanistAI, which are using AI in ways that genuinely empower participants to share and shape ideas at scale.

But I also wanted to zoom out and reflect on the moment we’re in. I’ve lived through more than a few hype cycles in the tech-for-good space, and it’s clear to me that we’re now at risk of repeating some old mistakes—this time with even higher stakes.

We need to be honest about how quickly powerful incumbents are learning to co-opt emerging technologies. And about the real dangers of replacing participation channels with synthesized participation, at a time where our democratic institutions cannot afford to lose any remaining political legitimacy. I believe AI can support democratic processes, but only if we use it to deepen and facilitate public involvement, not to simulate or replace it

Democracy has never been about engineering “optimal” policy outcomes—it’s about the contested, messy, and all-too-human process of reaching them together. Let’s make sure we don’t lose sight of that in the glare of our (admittedly) very shiny new toys.