Digital Democracy / AI · 2024 · Taiwan
Deliberative Democracy: AI for Inclusive Participation
Event Organizer & Moderator · AI Consensus
"One deliberation workshop. 600+ opinions. The bottleneck isn't participation — it's processing."
The Event
I organized and moderated a Fireside Discussion on AI and deliberative democracy, held within the vTaiwan and AI Consensus community. The session brought together panelists working at the intersection of civic technology, AI, and democratic participation.
I structured the entire session — designing the flow, preparing the presentation materials, and coordinating each panelist around the central question: can AI help make sense of a crowd without flattening it?
Taiwan has been a global leader in deliberative democracy — through vTaiwan, g0v, and structured public workshops that have shaped actual legislation since the 2014 Sunflower Movement. This event examined what that model looks like when AI enters the process, and whether it helps or harms the quality of democratic participation.
Topics
Broadcasting vs. Broad Listening
Traditional deliberation collects opinions one-way and summarizes them top-down. AI tools like Talk to the City (TTTC) and Pol.is enable "Broad Listening" — clustering and synthesizing citizen voices rather than simply tallying them. The discussion examined where this distinction matters most.
People-Centered Participation
AI can improve deliberative quality — but legitimacy requires human participation, not just efficiency. Algorithmic bias, digital divides, and data transparency are not edge cases. They are the design problem. The panel explored what "people-centered" AI governance actually means in practice.
Real-World Cases
The session drew on concrete examples: Taiwan's May 2024 protest (real-time opinion mapping via Pol.is QR codes in the streets), the Tokyo 2024 Governor election campaign of AI engineer Anno Takahiro (who used TTTC for broad-listening and received 154,638 votes as the #1 first-time candidate in his age group), and the OpenAI Democratic Inputs to AI grantee work — top 10 of 800+ applicants.
Lessons for Korea
The session explored what it would take to apply Taiwan's civic tech model in South Korea — and what prerequisites (civic infrastructure, institutional trust, open-source culture) must be in place before AI-assisted deliberation can scale.
Photos
With Peter Cui, Shawn Hung and Gisele Chou at the AI Consensus event, Taipei 2024.
The full panel in front of the session slide — Mashbean, Gisele Chou, Peter Cui, Taka Shunsuke, and Zoey Tseng.