2026
Cultivating Pluralism In Algorithmic Monoculture: The Community Alignment Dataset
ICLR 2026poster
How can large language models (LLMs) serve users with varying preferences that may conflict across cultural, political, or other dimensions? To advance this challenge, this paper establishes four key results. First, we demonstrate, through a large-scale multilingual human study with representative s…