Diffusion Mechanism

In the AI Story

The mechanism has specific implications for institutional change strategy in AI governance. The conventional approach — comprehensive national or international legislation — faces political barriers that typically delay implementation for years or decades. The diffusion approach begins at the level where political will is most accessible and institutional barriers are lowest: municipal AI Impact Assemblies, organizational Transition Deliberation Committees, school AI Governance Councils. Each successful implementation at any level provides the evidence base and practical experience that makes implementation at higher levels politically feasible.

The theory presupposes three factors operating together: demonstration effects (successful implementations produce documented outcomes), political entrepreneurship (specific actors recognize opportunities to mobilize evidence for implementation in new contexts), and crisis-driven demand (events that make inadequacy of existing arrangements impossible to ignore). The factors can operate independently but reinforce each other when aligned.

The application to AI governance is specific. The technological and social developments that constitute the AI transition are producing conditions favorable to crisis-driven demand: workforce displacement, educational disruption, informational ecosystem degradation, concentration of economic power. These conditions generate the political openings within which demonstration effects from pilot implementations can be converted into broader institutional change.

The model's limitation is that diffusion requires time — typically years even in favorable conditions — and AI deployment is proceeding faster than this timescale. Fung's response is that the structural governance decisions most amenable to participatory governance operate on timescales compatible with diffusion. Individual deployment decisions may require faster response, but the institutional frameworks within which deployment occurs can be built through diffusion even as deployment itself proceeds at AI timescales.

Origin

The theory emerged from Fung's analysis of how participatory budgeting spread from a single Brazilian city to hundreds worldwide. The observation that the diffusion pattern did not match conventional policy diffusion models — which emphasize political mobilization and coalition-building — drove the search for the specific mechanism that explained the observed pattern.

The theory draws on Everett Rogers' Diffusion of Innovations and on institutional-change literature in political science (particularly work on incremental institutional change by Kathleen Thelen and others). Fung's contribution was specifying how the general diffusion framework applies to participatory governance innovations and identifying the specific conditions that shape their spread.

Key Ideas

Demonstration drives diffusion. Documented success in one context generates demand for implementation in others, more reliably than coordinated political campaigns.

Start where barriers are lowest. Municipal and organizational implementations precede national legislation; successful lower-level implementations create political conditions for higher-level adoption.

Three factors operate together. Demonstration effects, political entrepreneurship, and crisis-driven demand reinforce each other when aligned.

Iterative refinement improves designs. Successive implementations learn from predecessors, producing institutional forms that improve through practice.