Accountable Human-AI Deliberation with LLMs: Scaling Collective Intelligence through Symbiotic Scaffolding
arXiv:2605.26940v1 Announce Type: new Abstract: Large language models (LLMs) can support democratic deliberation at scales previously constrained by turn-taking and facilitation bandwidth. Recent work shows that LLM-generated group statements are often preferred over human-mediated outputs, while theoretical analyses argue that LLMs relax the simultaneity constraints limiting collective intelligence. Yet pure LLM mediation risks collapsing pluralism, over-optimizing for agreement, and undermining legitimacy when participants cannot contest how they are represented. We propose a symbiotic human
The rapid advancement and adoption of large language models (LLMs) are pushing researchers to explore their capabilities beyond simple task automation, into complex social and political domains like deliberation.
This research addresses a critical challenge in scaling collective intelligence with AI, highlighting both the potential for enhanced participation and the risks of undermining legitimacy and pluralism if not carefully managed.
The proposed 'symbiotic scaffolding' approach suggests a path for integrating LLMs into human deliberation in a way that respects human agency and safeguards against AI's potential downsides in complex social interactions.
- · Platforms for online deliberation
- · Organizations requiring scalable consensus building
- · Democratic institutions seeking broader participation
- · Traditional facilitation services
- · Purely human-mediated deliberation processes (in terms of scale)
- · LLM developers ignoring ethical and social implications
Societies gain new methods for large-scale, efficient, and potentially more inclusive deliberation on complex issues.
The ethical and governance frameworks for AI's role in public discourse become increasingly critical and complex.
This could lead to new forms of democratic engagement and decision-making, potentially shifting power dynamics in policy formulation.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.CL