Democracy as a Scaled Collective Intelligence Process: Points of Vulnerability and Augmentation

The ideal democratic process aims to solve complex social decision problems, where diverse communities have potentially conflicting goals, through public deliberation, based on transparency, accountability, and trust. In practice, there are scaling limits to deliberative processes, both in terms of number of participants and cognitive complexity and democratic societies fall short of the ideal. Use of generative AI has been proposed to replace or augment the deliberative process. We argue that generative AI is an opaque process by nature, and hides issues with bias, power, accountability, and trust; and as such should not be directly involved in the decision-making process. We propose an alternate path, where decision-making is grounded in a fully transparent collective intelligence process, using a decision-oriented global structured knowledge base. Hybrid AI could help people approach and contribute to such a knowledge base, and watch over the coherence of expectations, actions and goals.