mlsanityData to 14 Oct 2026

Training-Free Cultural Alignment of Large Language Models via Persona Disagreement

arXiv 2605.10843 ↗First posted 11 May 2026cs.CL cs.AI cs.CYcited by 0

On the map: LLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment

Abstract (arXiv)

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.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Human values and AI feedback alignment, one of 10 directions in LLM alignment / RLHF / preference optimisation. Builds on 26 of its 43 references that are in the map. First posted on arXiv 11 May 2026.

Builds on

26 of 43 references.

Closest ancestors · 5

Same direction, newest first

  1. 2025Dropouts in Confidence: Moral Uncertainty in Human-LLM Alignment17 Nov
  2. 2025Exploring Persona-dependent LLM Alignment for the Moral Machine Experiment15 Apr
  3. 2024Language Model Alignment in Multilingual Trolley Problems2 Jul
  4. 2024The PRISM Alignment Dataset: What Participatory, Representative and Individualised Human Feedback Reveals About the Subjective and Multicultural Alignment of Large Language Models24 Apr
  5. 2024A Roadmap to Pluralistic Alignment7 Feb

Other references · 21

  1. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,708
  2. 2017On Calibration of Modern Neural Networkscited by 4,960
  3. 2015Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learningcited by 4,448
  4. 2023Improving Factuality and Reasoning in Language Models through Multiagent Debatecited by 1,784
  5. 2023Representation Engineering: A Top-Down Approach to AI Transparencycited by 1,336
  6. 2024Refusal in Language Models Is Mediated by a Single Directioncited by 1,032
  7. 2023Encouraging Divergent Thinking in Large Language Models through Multi-Agent Debatecited by 1,011
  8. 2023Whose Opinions Do Language Models Reflect?cited by 773
  9. 2018Reinforcement Learning and Control as Probabilistic Inference: Tutorial and Reviewcited by 482
  10. 2024BLEnD: A Benchmark for LLMs on Everyday Knowledge in Diverse Cultures and Languagescited by 173
  11. 2023Controlled Decoding from Language Modelscited by 135
  12. 2024ARGS: Alignment as Reward-Guided Searchcited by 103
9 more
  1. 2017Information Theoretic Model Predictive Control: Theory and Applications to Autonomous Drivingcited by 90
  2. 2024Unintended Impacts of LLM Alignment on Global Representationcited by 66
  3. 2025Randomness, Not Representation: The Unreliability of Evaluating Cultural Alignment in LLMscited by 54
  4. 2023The Moral Machine Experiment on Large Language Modelscited by 25
  5. 2024Large-scale moral machine experiment on large language modelscited by 9
  6. 2025No Free Lunch in Language Model Bias Mitigation? Targeted Bias Reduction Can Exacerbate Unmitigated LLM Biasescited by 7
  7. 2025CAReDiO: Cultural Alignment via Representativeness and Distinctiveness Guided Data Optimizationcited by 6
  8. 2026Culturally Grounded Personas in Large Language Models: Characterization and Alignment with Socio-Psychological Value Frameworkscited by 3
  9. 2026Mind the Gap in Cultural Alignment: Task-Aware Culture Management for Large Language Modelscited by 3

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Whose Norms? Disentangling Cultural and Personal Alignment in Large Language Models
  2. 2026ACE-Align: Attribute Causal Effect Alignment for Cultural Values under Varying Persona Granularities ↗
  3. 2026Cultural Value Alignment Via Latent Activation Steering in Large Language Models
  4. 2026Aligned but Flattened: Analyzing the Trade-off between Cultural Alignment and Diversity in LLMs
  5. 2026Scenario-based Probing and Steering Cultural Values in Large Language Models--Extended Version
  6. 2023Cultural Bias and Cultural Alignment of Large Language Models
  7. 2025Multiple LLM Agents Debate for Equitable Cultural Alignment
  8. 2025Cultivating Pluralism In Algorithmic Monoculture: The Community Alignment Dataset