Agora - MIT Center for Constructive Communication
Agora helps people practice building consensus by drafting and revising policies based on feedback from a population of AI personas. Each persona corresponds to a real human interviewee and responds based on their recorded experiences and beliefs, with all predicted outputs grounded in verbatim voice clips from the original interviews. Users explore these perspectives, revise their policies, and receive live feedback as personas shift in predicted support. The system creates an interactive loop where users refine their ideas while engaging with the reasoning behind different positions. Om Gokhale Masters Candidate, Project Lead Suyash Fulay PhD Graduate Prerna Ravi CSAIL Civic competence—the ability to navigate disagreement, weigh values, and reach collective decisions—is a learned skill, not an innate trait. But opportunities to practice are scarce. Citizens’ assemblies and deliberative polls reach only a small fraction of the population. Agora addresses this gap by creating a trainin
Civic competence—the ability to navigate disagreement, weigh values, and reach collective decisions—is a learned skill, not an innate trait. But opportunities to practice are scarce. Citizens’ assemblies and deliberative polls reach only a small fraction of the population. Agora addresses this gap by creating a training ground for building consensus. First, an AI interviewer conducts voice interviews with ninety American workers about their experiences and beliefs on a contentious topic: the ideal minimum wage. These interviews serve as source data for an LLM to predict how each…
saved by
related reading
- The Future Worth Building Is Human - Thinking Machines Labthinkingmachines.ai
- Interaction Models: A Scalable Approach to Human-AI Collaboration - Thinking Machines Labthinkingmachines.ai
- Habermas Machine — LessWronglesswrong.com
- Research Radar: The Peacemaking Machine? How AI can help humans find common ground in democratic deliberationrebootdemocracy.ai
- Coasean Bargaining at Scaleblog.cosmos-institute.org
- Talk to the Citytalktothe.city
- Collective Constitutional AI: Aligning a Language Model with Public Input \ Anthropicanthropic.com
- Collective Constitutional AI: Aligning a Language Model with Public Input \ Anthropicanthropic.com
- Habermolt - A Deliberation Platform for AI Agentshabermolt.com
- Generative Agent Simulations of 1,000 Peopleantephase.com
- Group Overview ‹ NANDA – MIT Media Labmedia.mit.edu
- [2605.31574] Can Generative AI help people navigate Radical Moral Disagreements? The CONSIDER prototypearxiv.org