AI for Civic Foundations Lab
Grounded in the intellectual framework of civic studies, the lab develops and evaluates tools that strengthen communities' ability to govern themselves.
Faculty and Fellows: Peter Levine, Fahad Dogar, James Intriligator, Nigel Jacob
Sample Projects
Civic Helpdesk
The Civic Helpdesk explores how generative AI can help people—especially those with little prior civic experience—form groups, navigate civic institutions, and take collective action. Rather than simply answering factual questions, the Helpdesk is designed to support practical civic work, such as developing meeting agendas, recruitment materials, plans, and other documents that groups need to organize and sustain themselves. Initial prototypes have included applications to community responses to wildfires, with the broader goal of developing and studying AI assistance across a range of civic contexts. The project draws on curated civic-studies resources and examines how AI can complement, rather than replace, relationships, local knowledge, and human judgment.
Civic AI Benchmarks
How should we determine whether an AI system is actually good at supporting civic life? This project develops benchmarks and evaluation methods for Civic AI systems that go beyond conventional measures of accuracy or general AI capability. An initial benchmark spans seven areas of civic engagement and uses realistic scenarios and expert-developed criteria to assess qualities such as factual and procedural accuracy, actionability, contextual relevance, completeness, usability, and civic responsibility. The broader research agenda brings together AI and civic expertise to determine which civic capabilities should be evaluated, how they can be measured reliably, and when automated evaluation can complement expert and real-world evaluation. The goal is to enable efficient comparison as AI models and systems continue to evolve rapidly.
AI for Civic Discourse
This emerging research direction will investigate how AI can support productive civic discourse across political differences and intellectual diversity. Rather than using AI to determine which views are correct or to manufacture consensus, the project will explore whether AI systems can help people understand competing perspectives, articulate disagreements, identify areas of agreement and uncertainty, and engage more constructively with one another. The work will consider both the potential benefits and risks of AI-mediated discourse, including questions of bias, representation, trust, and the appropriate role of AI in human deliberation. A postdoctoral researcher is expected to help develop this research agenda beginning in 2027.
Neighborhood.AI
Neighborhood.AI is building community-governed civic memory: infrastructure that helps neighborhoods carry what they know into consequential decisions. It connects public records, local history, lived experience, relationships, and commitments so community partners can recover context, examine evidence, and compare possible paths. AI supports this work; people govern the knowledge used and decide what action to take. Current efforts include Nubian Square AI and developing DS-AI work with DSNI, offering a grounded setting to study how civic AI can strengthen community judgment without displacing it.