CoCoA: Context-Conditional Cultural Alignment for Large Language Models

arXiv:2608.29492v1 Announce Type: new Abstract: Large Language Models (LLMs) often favor Western-associated entities across cultural contexts. Conventional debiasing methods aim for uniform neutrality, but cultural bias mitigation demands context-conditional behavior, preferring culturally appropriate entities when cultural cues are present and remaining neutral when they are absent. We propose CoCoA (Context-Conditional Cultural Alignment), a framework that learns this behavior through dual-context training on the same entity pairs under contexts with and without cultural cues. CoCoA combines a contrastive alignment objective with calibration and drift regularization, optimized through goal-aware gradient reconciliation. We evaluate CoCoA on CAMeL and Camellia, two entity-centric cultural bias benchmarks, across ten language settings and four LLMs. CoCoA reduces the Cultural Bias Score from 43 to 24 on average while maintaining near-neutral preferences at 50.2, with minimal impact on general performance across five standard benchmarks. These findings highlight that effective cultural alignment requires context-conditional modeling rather than uniform debiasing, and establish a new direction for mitigating entity-centric cultural bias in LLMs.

Related

• Scenario-based Probing and Steering Cultural Values in Large Language Models--Extended Version
• Progressing beyond Art Masterpieces or Touristic Cliches: how to assess your LLMs for cultural alignment?
• Cultural Value Alignment Via Latent Activation Steering in Large Language Models
• C-Mining: Unsupervised Discovery of Seeds for Cultural Data Synthesis via Geometric Misalignment

Source: arXiv cs.CL | 2026-09-01