Legitimation Mechanism

In the AI Story

The concept extends traditional analyses of institutional legitimacy (Weber, Habermas) to the specific forms legitimation takes in technology governance. The distinctive feature of AI-governance legitimation is that it exploits complexity: citizens evaluating whether governance is adequate typically cannot assess the substance of expert advisory processes, so institutional signals (Harvard academics, former government officials, civil society representatives) substitute for substantive evaluation.

The mechanism compounds the fishbowl condition by making exclusion appear to be inclusion. The affected populations not merely excluded from decision-making but presented with evidence that governance is adequate — evidence that makes mobilization for substantive inclusion harder because it suggests the problem is already being addressed.

The distinction between legitimation mechanisms and genuine participatory governance is operational, not merely formal. Both involve meetings, discussions, reports, and advisory bodies. The difference lies in whether the three conditions of empowered participatory governance are substantively satisfied. A board that lacks binding authority, operates without accessibility for affected populations, or fails to engage in structured deliberation is a legitimation mechanism regardless of the credentials of its members.

The distinction applies beyond corporate ethics boards to many forms of AI governance: regulatory comment periods that process inputs without converting them into outputs, congressional hearings that display concern without producing action, academic-industry partnerships that generate publications without generating change. Each can function as legitimation regardless of participants' sincerity, because the mechanism's effect depends on institutional structure rather than individual intent.

Origin

The concept emerges from the sociological tradition of analyzing institutional legitimacy, particularly Max Weber's typology of legitimate authority and Habermas's work on the legitimation crisis of late capitalism. Fung's contribution is extending this analysis to the specific institutional forms that legitimation takes in contemporary technology governance.

The empirical basis for the concept includes Fung's own analysis of corporate governance structures, academic work on AI ethics boards (particularly by Kate Crawford and Meredith Whittaker on the limitations of existing corporate AI ethics processes), and broader sociological work on how technical complexity enables institutional legitimation strategies.

Key Ideas

Complexity enables legitimation. Citizens who cannot evaluate substantive governance rely on institutional signals, which legitimation mechanisms provide without governance substance.

Advisory authority is the critical deficit. The most common legitimation failure is advisory rather than binding authority — a structural feature that ensures recommendations can be disregarded without consequence.

Exclusion appears as inclusion. The mechanism's distinctive damage is that it makes affected populations appear represented in processes from which they are structurally excluded.

The distinction is operational, not formal. Genuine participation and legitimation mechanisms share surface features; they differ in whether the three conditions are substantively satisfied.