Civic AI

“Civic AI” is the use of artificial intelligence in the service of democracy and civic life — for participation, governance, advocacy, public-interest media, and the public good. It’s a fast-growing slice of civic tech: the Civic Tech Field Guide alone has catalogued over 700 civic-AI projects, organisations, and resources.

How the field is mapped

There’s no single agreed taxonomy yet, but practitioners tend to organise civic AI two ways at once — by domain and by type.

By domain (the pillars the Civic Tech Field Guide uses for civic tech generally, now shot through with AI):

• Engagement & participation — platforms that help people deliberate, decide, and shape policy together
• Gov tech — especially tools that shift power and open up decision-making inside government
• Open government & civic data — making public data usable
• Advocacy — political campaigns and watchdog / accountability work
• Media — journalism, games for change, and fighting disinformation
• Emerging tech — AI alongside drones, satellites, and VR

By type — is it open-source AI? does it involve governance? is it a bot, or a training dataset (for example, getting low-resource languages better represented in AI models)?

Underneath all of it sit foundational layers: an AI tool for civic life is only worth using if it’s secure, privacy-protecting, fair, and accessible — which is often not the case. So “responsible AI” isn’t a separate world from civic AI; it’s the floor.

Why a taxonomy at all

Mapping the field lets builders learn from each other, helps people find tools that already exist, and lets funders and researchers compare like with like. The same impulse drives this wiki’s toolkit and run reports.

A useful reframe: AI as a social technology

Before mapping where civic AI shows up, it helps to be clear about what AI is — because the framing you start from quietly shapes the politics. Henry Farrell and Cosma Shalizi argue that a large language model is best understood not as a nascent autonomous agent, the “Singularity” inherited from 1990s science fiction, but as a social technology: “a systematic means of reorganizing social relationships among human beings.” On that view AI sits in a lineage with other social technologies — “institutions of governance such as bureaucracies, markets, and even democracy” — each of them a way of compressing a complex world into something a society can act on. Their conclusion is deflating and useful: “we should treat AI as a new social technology which will alleviate some problems, exacerbate others, and create new ones, just as other social technologies have done in the past.”

Why this matters for civic AI: if AI is an agent on the verge of thinking for us, the civic task is to contain or align it. If it’s a social technology that reorganises relationships, the task is the familiar democratic one — deciding whose relationships it reorganises, in whose interest, and under whose governance. It’s the same instinct the solidarity-economy writers reach for when they reject “the notion of artificial intelligence, which implies a magical, autonomous force” and rename it collective intelligence (see democratising AI), and the one behind Audrey Tang’s plurality. It also sets up the sharper fight over whether AI can stand in for the work bureaucracies and courts do — see does AI weaken democratic institutions?

Sources

• Matt Stempeck (Civic Tech Field Guide), “Civic AI” — Metagov seminar, 2025: youtube.com/watch?v=SonfdR0_h2w
• Henry Farrell & Cosma Rohilla Shalizi, “AI as Social Technology” — Knight First Amendment Institute, 2026: knightcolumbia.org. Builds on Farrell, Gopnik, Shalizi & Evans, “Large AI models are cultural and social technologies,” Science, 2025.
• Civic Tech Field Guide