Plurals: A System for Guiding LLMs Via Simulated Social Ensembles
This addresses viewpoint bias in LLMs for AI developers and researchers, though it appears incremental as it builds on existing deliberation concepts.
The authors tackled the problem of language models favoring certain viewpoints by introducing Plurals, a system for pluralistic AI deliberation using simulated social ensembles, which in experiments produced output resonant with relevant audiences in 75% of trials compared to zero-shot generation.
Recent debates raised concerns that language models may favor certain viewpoints. But what if the solution is not to aim for a 'view from nowhere' but rather to leverage different viewpoints? We introduce Plurals, a system and Python library for pluralistic AI deliberation. Plurals consists of Agents (LLMs, optionally with personas) which deliberate within customizable Structures, with Moderators overseeing deliberation. Plurals is a generator of simulated social ensembles. Plurals integrates with government datasets to create nationally representative personas, includes deliberation templates inspired by deliberative democracy, and allows users to customize both information-sharing structures and deliberation behavior within Structures. Six case studies demonstrate fidelity to theoretical constructs and efficacy. Three randomized experiments show simulated focus groups produced output resonant with an online sample of the relevant audiences (chosen over zero-shot generation in 75% of trials). Plurals is both a paradigm and a concrete system for pluralistic AI. The Plurals library is available at https://github.com/josh-ashkinaze/plurals and will be continually updated.