Hyperdemocracy: Towards Creative Consensus Building between Humans and AI

Effective consensus building is a critical challenge in diverse domains, including international disputes, societal polarization, and organizational decision-making. This research project envisions a future termed 'Hyperdemocracy,' where a multitude of AI agents collaborate with humans to support consensus building while fostering democratic deliberation. These AI agents, based on Large Language Models (lLMs), can introduce diverse perspectives through text-based dialogue, guiding discussions toward mutual understanding. Advances in LLMs offer a promising avenue for mitigating persistent issues in group decision-making, such as groupthink and cognitive biases. Field experiments conducted in Afghanistan and Indonesia have demonstrated that our multi-agent system can stimulate deliberation and promote opinion change among participants, even in highly fragmented contexts. This paper details the system's design philosophy, presents the results of its real-world validation, and discusses the ethical considerations and future prospects. We suggest that this technology holds transformative potential for conflict resolution and fostering inclusive dialogue.