Building democratic capability
We work with democratic practitioners, facilitators, process convenors, researchers, tool builders, funders, and governments to work towards an ecosystem that can support high-quality democratic systems + decision processes.
Process Pattern Language and Deliberative Workflows A standardized pattern language for the design and execution of collective decision-making processes, to help knowledge about best practices accumulate and transfer across contexts.
Research Directions: AI for Deliberative Democracy A set of open research questions that direct new AI capabilities at existing challenges in deliberative democracy, such as scenario testing, multi-modal learning, and facilitator information management.
Deliberative Tools Access Program Access to deliberative tools for organizations and independent researchers to experiment, learn, and develop a progressively more robust ecosystem of deliberative tools.
Facilitator Resource Sharing Groups Coordinating intimate groups of world-class deliberation facilitators and practitioners to facilitate high-trust knowledge sharing, develop best practices, and identify critical missing infrastructure and tools.
Realizing democratic AI
Decisions about how AI transforms society should not be made unilaterally. We work with frontier AI companies, governments, researchers, and civil society to drive the design and adoption of meaningfully democratic systems for AI alignment + governance.
Democracy Levels A framework for evaluating the degree to which decisions in a given domain are made democratically. It provides a common language for milestones relevant to the goals of the democratic AI, pluralistic AI, participatory AI, and public AI ecosystems.
Safeguarded AI For ARIA, designing and building processes for deliberatively defining formal safety specifications for advanced AI systems, and developing legal structures for AI organizations that would allow them to commit to acting in accordance with such collective decisions.
Case Law for AI Exploring how "case law" approaches inspired by judicial practices may provide a more effective knowledge representation and participant experience for connecting deliberative public input into AI alignment and inference systems.