Training-Free Cultural Alignment of Large Language Models via Persona Disagreement
For developers of LLMs serving global users, this provides a training-free, black-box method to align models with diverse cultural moral preferences, addressing a known bottleneck in cultural fairness.
DISCA reduces cultural misalignment of LLMs across 20 countries by 10-24% on MultiTP and 2-7% on open-ended scenarios, using inference-time logit correction from World Values Survey personas, without fine-tuning or white-box access.
Large language models increasingly mediate decisions that turn on moral judgement, yet a growing body of evidence shows that their implicit preferences are not culturally neutral. Existing cultural alignment methods either require per-country preference data and fine-tuning budgets or assume white-box access to model internals that commercial APIs do not expose. In this work, we focus on this realistic black-box, public-data-only regime and observe that within-country sociodemographic disagreement, not consensus, is the primary steering signal. We introduce DISCA (Disagreement-Informed Steering for Cultural Alignment), an inference-time method that instantiates each country as a panel of World-Values-Survey-grounded persona agents and converts their disagreement into a bounded, loss-averse logit correction. Across 20 countries and 7 open-weight backbones (2B--70B), DISCA reduces cultural misalignment on MultiTP by 10--24% on the six backbones >=3.8B, and 2--7% on open-ended scenarios, without changing any weights. Our results suggest that inference-time calibration is a scalable alternative to fine-tuning for serving the long tail of global moral preferences.