Synthetic Chamber: Constitutionalising Algorithmic Representation Through Deliberative Data and Multi-Agent Reason-Giving

This paper responds to the deepening crisis of representative democracy—the hollowing of political parties, the procedural decay of parliaments, and the misframing of who and what can be represented—by proposing the synthetic chamber, a democratic innovation and constitutional institution that re-engineers representation through the use of artificial intelligence. Whereas existing systems tend to reduce AI to administrative assistance or electoral campaigning, the synthetic chamber reconceives it as a deliberative organ designed to repair the broken circuits between citizens, institutions, and the long-term or non-human constituencies currently excluded from the political. Building on Bruno Latour’s Parliament of Things and Christopher Stone’s guardianship model, the chamber extends legal agency and accountability beyond the human, offering a framework in which AI acts as a transparent mediator rather than an opaque oracle. Highlighting research-creation practices with political AI, the article demonstrates how multi-agent deliberation can institutionalise contestation rather than consensus. Multi-agent systems for large language models can generate plural, auditable opinions that legislatures could publicly address. The result is a new constitutional medium that counters the algorithmic capture of democratic voice by reinstating deliberation, auditability, and agonism as core mechanisms of democratic legitimacy, moving the discussion of AI from a symptom of political exhaustion into an instrument of constitutional renewal.