Technical Constitution

In the AI Story

The concept allows Winner's framework to address the AI moment with theoretical precision. The transition from pre-AI to AI-mediated knowledge work is not merely a productivity enhancement. It is a restructuring of what capabilities exist in the society, who possesses them, at what cost, on what terms. These are constitutional questions.

A constitutional moment, in democratic theory, requires the informed consent of the governed. It requires public deliberation about the kind of society the new constitution will produce. It requires the participation of populations whose lives will be structured by its terms. What it does not permit — what is antithetical to constitutional governance — is the assumption that the rewriting is a natural process to be accommodated rather than a political process to be shaped.

The technical constitution concept exposes the inadequacy of the current AI governance conversation. The EU AI Act, American executive orders, and similar frameworks operate as ordinary legislation — rules within an existing constitutional framework. But the AI transition is rewriting the framework itself. Ordinary legislation is inadequate to constitutional moments; the response must be constitutional in scope.

Constitutional moments have occurred before. The New Deal was one — a democratic rewriting of the economic constitution in response to market failure at civilizational scale. The civil rights era was another. Each involved not merely new laws but new institutional structures, new democratic relationships, new forms of participation. The AI moment demands comparable ambition.

Origin

The concept is implicit throughout Winner's work, particularly in Autonomous Technology and The Whale and the Reactor, where the question of democratic governance of technological systems is persistently framed in constitutional language. Eric Deibel's 2025 paper made the concept explicit as part of a broader effort to extend Winner's framework into AI governance debates.

The parallel to Bruce Ackerman's theory of constitutional moments in American political history is deliberate. Ackerman distinguished between ordinary politics (legislation within existing constitutional frameworks) and constitutional politics (the rewriting of the frameworks themselves). Applied to technology, Winner's framework asks whether democratic societies can conduct technical-constitutional politics or whether they will continue to treat each technological transition as ordinary.

Key Ideas

Two constitutions, not one. Political and technical constitutions operate in parallel, each distributing authority, each shaping what the other can do.

Constitutional moments require participation. The rewriting of fundamental rules — political or technical — is the paradigmatic occasion for democratic engagement; its absence is democratic failure.

Ordinary regulation is inadequate. The EU AI Act and comparable frameworks treat AI as a regulated industry; the framework demands treating it as a constitutional rewriting.

Capabilities as constitutional matter. What capabilities exist, who has them, on what terms — these are constitutional questions, not market questions.

Democratic rewriting is possible. The New Deal, the civil rights era, the labor movement of the early twentieth century all demonstrate that constitutional moments can be conducted democratically when institutions commit to it.

Debates & Critiques

Critics argue that 'constitution' is the wrong metaphor for technical infrastructure — that the historical connotations of constitutional politics (national foundings, revolutionary moments) do not apply to the gradual accretion of technological capability. Defenders reply that the gradualness is the problem: the slow pace of change has allowed civilizational transformation to proceed without the democratic engagement that constitutional politics demands, and that recovering the constitutional frame is the prerequisite for recovering the engagement.