PERSONA: Evaluating Pluralistic Alignment in LLMs

Our approach utilizes synthetic personas, crafted through a combination of US census data and procedural generation, to simulate a wide array of user profiles with diverse demographic and idiosyncratic attributes. We present a detailed methodology for constructing a representative demographic of 1,586 personas, each enriched with individualistic personality traits and core values. Leveraging this synthetic demographic, we generate a large-scale preference dataset containing 3,868 prompts and 317,200 pairs of diverse feedback.

This dataset enables the evaluation of language models' ability to align with both group-level and individual preferences across various controversial and value-laden topics. Our contributions include a systematic evaluation of current LM capabilities in role-playing diverse users, verified through human judges, and the establishment of a benchmark for pluralistic alignment approaches. Our work aims to facilitate the development of more inclusive and representative language models, paving the way for future research in global pluralistic alignment.

• Synthetic Personas: A novel methodology for creating diverse, representative personas using US census data and procedural generation.
• Large-scale Dataset: 3,868 prompts and 317,200 preference pairs capturing diverse viewpoints.
• Evaluation Framework: Systematic evaluation of LM capabilities in role-playing diverse users.
• Benchmark: Establishment of a reproducible benchmark for pluralistic alignment approaches.