Scaling Up Good Listening: An Assessment Framework for AI-Powered Mass Deliberation Models

The challenges of scaling up deliberative processes to mass audiences have long been highlighted by deliberative theorists. Apart from the difficulty of keeping content quality at a high enough level as more and more people get involved, the technical feasibility of mass participation in a structured form of deliberation has been a serious constraint. The development of digital platforms and AI systems are now making it technically possible to extend structured participation to wider audiences. This article addresses the following question: How can we ensure that good listening is scaled up in these new contexts? Drawing on an analytical framework based on recent contributions in the areas of deliberative democracy and AI, we evaluate the ability of current models of AI-powered mass deliberation to incentivize receptive, responsive, and apophatic<em> </em>listening. We further develop an assessment tool, the “Listening Incentives Score,” that can be used to assess whether AI-powered mass deliberation models provide participants with the adequate channels, facilitation, training, and systems of rewards and sanctions to incentivize them to engage in good listening.