Beyond the Habermas machine: rethinking the democratic meaning of AI-assisted deliberation

Advances in artificial intelligence have generated renewed interest in using computational systems to facilitate democratic deliberation at scale. While recent efforts, exemplified by the ‘Habermas Machine,’ promise efficiency, balance, and consensus, they raise fundamental questions about the meaning of deliberation itself. This article offers a theoretical critique of deliberation using AI, focusing on systems that operationalize deliberation as a consensus-optimizing procedure. Drawing on Habermasian theories of communicative rationality, deliberative systems scholarship, and distinctions between private and public deliberation, the article argues that current AI approaches risk translating deliberation from communicative rationality into computational rationality. This shift privileges public acceptability and formal balance over critical contestation, reflexivity, and democratic legitimacy. In response, the article reconceptualizes AI-mediated communication by foregrounding private, informal interaction as a ‘pre-public’ space within the deliberative system, where citizens can explore disagreement in judgement-free context and develop deliberative capacities.