Synthocracy: Citizens’ Assemblies, Public Consultation, and Collective Intelligence
What Happens When AI Starts Co-Deciding? The Quiet Shift from Intelligence to Power
Synthocracy is a decision order in which humans formally continue to govern, manage, vote, approve, or take responsibility, while the real work of detecting, filtering, prioritizing, recommending, classifying, and sometimes executing decisions increasingly passes through AI systems, predictive models, agents, data infrastructures, and digital platforms.
Martin Novak, Novakian Paradigm Institute
Citizens’ Assemblies, Public Consultation, and Collective Intelligence
The democratic promise of AI becomes most concrete when we move from abstract debate to actual participatory processes. Citizens’ assemblies, public consultations, participatory budgeting, municipal planning, climate policy consultations, regulation feedback, and collective intelligence platforms all face the same practical challenge: democracy produces more voices than institutions can easily process. People submit comments, proposals, objections, stories, local knowledge, technical concerns, emotional testimony, amendments, petitions, survey responses, meeting transcripts, and open-ended feedback. The difficulty is not that the public has nothing to say. The difficulty is that public input often arrives in forms too large, fragmented, uneven, and complex for ordinary institutions to understand well.
This is one of the reasons participation can become symbolic. A government announces a consultation, opens a portal, receives thousands of responses, publishes a summary, and claims that citizens were heard. But what does “heard” mean? Were the responses read carefully? Were minority concerns preserved? Were repeated arguments counted as public weight or treated as duplication? Were emotional testimonies considered evidence of lived experience or dismissed as anecdotal? Were technical submissions given too much influence because they sounded professional? Were poorly written but important concerns overlooked? Were organized campaigns separated from individual voices? Were citizens able to see how their input changed the final decision? Without a serious method for processing public input, participation risks becoming ritual.
AI can help here. It can translate documents, summarize proposals, classify responses, identify themes, detect recurring concerns, compare arguments, group similar submissions, highlight minority positions, produce accessible briefings, and help officials process large volumes of citizen input. It can support deliberation before, during, and after public participation. Before a process begins, AI can help explain the issue in plain language. During the process, it can help citizens navigate options and formulate their concerns. After the process, it can help institutions map what was said, where the conflicts lie, and which points require further response.
Consider a citizens’ assembly on climate policy. Participants may need to understand energy systems, emissions targets, household costs, industrial transition, transport, agriculture, taxation, jobs, regional inequality, and long-term risk. The material can be overwhelming. AI could support the process by generating layered explanations: a short orientation for first contact, a more detailed explanation for deeper study, and technical summaries for participants who want to inspect assumptions. It could compare policy options, show likely trade-offs, summarize expert testimony, and identify questions that remain unresolved. It could help participants see that disagreement may concern not only facts, but also values: fairness, speed, burden-sharing, intergenerational responsibility, and trust in implementation.
Consider municipal planning. A city may consult residents about transport changes, housing density, green spaces, school zones, parking, noise, safety, local commerce, cycling infrastructure, public transport routes, or flood protection. Citizens often know things that planners miss: where children actually cross the street, which bus connection fails in practice, where elderly residents feel unsafe, which small shop depends on short-term parking, which area floods after heavy rain, which proposed route looks efficient on paper but ignores local habits. AI could help group these observations by location, issue, urgency, and affected population. It could create a map of concerns rather than a pile of comments. It could help officials see patterns without reducing every citizen to a data point.
Consider participatory budgeting. Residents propose projects: playgrounds, benches, lighting, crossings, trees, community centers, sports facilities, senior services, accessibility improvements, local safety measures, cultural events, or small infrastructure upgrades. AI could help merge duplicate proposals, clarify cost categories, identify projects serving similar needs, translate technical requirements, and make comparisons easier for voters. It could also help citizens understand why a proposal may be feasible, too expensive, legally difficult, or dependent on another authority. Used well, AI could make participatory budgeting less confusing and more transparent.
Consider regulation feedback. When a government consults businesses, civil society, professionals, experts, citizens, and affected communities on a new rule, the volume of submissions can become enormous. AI could help classify comments by article, concern, sector, legal issue, cost impact, rights impact, and proposed amendment. It could identify where many stakeholders raise the same problem and where a small group raises a serious but less visible issue. It could help public officials avoid readin