A Very Brief Introduction to Public AI

This short course introduces public AI as an approach to treating AI as public infrastructure. It emphasizes political economy, institutional design, and practical implementation rather than abstract ethics or purely technical alignment.

Each week combines shared readings with a concrete artifact that can be produced individually or collectively.

Week 1 — Foundations: What Is Public AI?

Core question

What does it mean to treat AI as public infrastructure, and how has this idea emerged across different institutional contexts?

Readings

Selected excerpts from:

Discussion prompts

  • What core features of public AI recur across these papers? Where do they diverge?
  • Is public AI framed more as infrastructure, policy intervention, or counterpower?

Applied exercise (output)

Comparative synthesis memo (≈1 page)
Identify 3–5 shared claims about public AI across the readings and 1–2 substantive disagreements. Distinguish what appears settled from what remains contested.

Week 2 — Why Public AI? Political Economy, Power, and Public Options

Core question

Why is public AI needed given existing tools like regulation, antitrust, and public procurement?

Readings

Optional readings

  • The Labor Market Impacts of Technological Change — David Autor (2024)

Background reference

  • Anthropic Economic Index Report (2026) — Anthropic

Discussion prompts

  • What can public options do that regulation or antitrust alone cannot?
  • Does public AI complement antimonopoly approaches, or risk weakening them?

Applied exercise (output)

Problem-framing memo (≈1 page)
Choose one political–economic justification for public AI (e.g. monopoly power, democratic legitimacy, state capacity). Argue why regulation or procurement alone is insufficient.

Optional readings

Week 3 — National Strategies, Sovereignty, and Coordination

Core question

How does public AI relate to national AI strategies, sovereignty claims, and multilateral coordination?

Readings

Discussion prompts

  • When does sovereignty reinforce public accountability, and when does it undermine it?
  • Is the Airbus analogy helpful for thinking about public AI, or misleading?

Applied exercise (output)

Strategy comparison brief (diagram + short text)
Compare two approaches (e.g. national public AI vs multilateral public AI). Identify coordination gains, governance risks, and failure modes.

Week 4 — Building Public AI Locally

Core question

What are realistic ways to build, pilot, or support public AI at local or institutional scales?

Readings

  • Public AI Libraries — initiative overview
  • Minimum Viable Public AI Utility — Public AI Network (link forthcoming)

Optional readings

Discussion prompts

  • Where do local institutions (libraries, universities, municipalities) have real leverage?
  • What makes a public AI initiative durable rather than symbolic?

Applied exercise (output)

Local action plan (1–2 pages)
Identify one plausible intervention (event, partnership, pilot proposal, policy memo). Specify the target institution, goals, allies, timeline, and success criteria.

Glossary

public AI (the idea)
An approach to treating AI as public infrastructure, emphasizing democratic governance, broad accessibility, and accountability to the communities that AI systems serve. Distinct from both purely private AI and state-owned AI systems.

Public AI (the movement)
The global movement of researchers, practitioners, policymakers, and communities working to realize public AI as infrastructure for the common good. Sometimes abbreviated as Public AI.

Public AI Network
An open community within the Public AI movement that curates resources like this handbook, coordinates research and advocacy, and maintains publicai.network as a hub for the movement.

Public AI Inference Utility
A particular product developed within the Public AI movement: an AI inference service operated as a public utility, with governance structures ensuring public accountability, sustainable financing, and service obligations to communities.

Projects and Initiatives

Apertus
A public AI project developing open infrastructure and tools for democratic AI governance and deployment.

NDIF (National Deep Inference Fabric)
A proposed framework for distributed public AI infrastructure that enables secure, privacy-preserving AI inference across institutional boundaries while maintaining public accountability.

Public AI Libraries Project
An initiative to position public libraries as access points and stewards of public AI services, building on libraries’ existing role as democratic institutions providing equitable access to information and technology.

Public option
A publicly provided alternative to private services that competes alongside private providers while pursuing public goals. In the AI context, offers choice while setting standards for quality, access, and accountability.

Public infrastructure
Essential systems and services that support collective life and economic activity, traditionally including roads, water, electricity, and increasingly digital systems. Public AI positions AI systems as infrastructure requiring similar governance and access principles.

AI sovereignty
The capacity of nations or regions to develop, deploy, and govern AI systems according to their own values and interests, rather than depending entirely on systems controlled by other jurisdictions or private corporations.

Mechanistic interpretability
Technical approaches to understanding how AI systems actually work internally, enabling better governance, debugging, and accountability of AI systems used in public contexts.