Conversational Swarm Intelligence (CSI) Enhances Groupwise Deliberation
Louis Rosenberg, Gregg Willcox, Hans Schumann, Ganesh Mani

TL;DR
Conversational Swarm Intelligence (CSI) leverages LLMs to enable large groups to conduct effective, balanced, and scalable real-time deliberations, outperforming standard chat in participation and user satisfaction.
Contribution
This paper introduces CSI, a novel system that combines small-scale deliberative reasoning with large-scale collective intelligence using LLMs, enhancing group deliberation scalability and effectiveness.
Findings
CSI increased content contribution by 51% (p<0.001).
CSI resulted in 37% less disparity in participation.
Participants preferred CSI and felt more impactful.
Abstract
Real-time conversational deliberation is a critical groupwise method for reaching decisions, solving problems, evaluating priorities, generating ideas, and producing insights. Unfortunately, real-time conversations are difficult to scale, losing effectiveness as groups grow above 5 to 7 members. Conversational Swarm Intelligence (CSI) is a new technology modeled on the dynamics of biological swarms. It aims to enable networked groups of any size to hold productive real-time deliberations that converge on unified solutions. CSI leverages the power of Large Language Models (LLMs) in a unique and powerful way, allowing real-time dialog among small local groups while simultaneously enabling efficient content propagation across much larger populations. In this way, CSI combines the benefits of small-scale deliberative reasoning and large-scale collective intelligence. In this study, we…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsTeam Dynamics and Performance · Expert finding and Q&A systems · Knowledge Management and Sharing
