Cultural Value Alignment Via Latent Activation Steering in Large Language Models

2026-05-25

AI Summary

Large language models commonly exhibit cultural homogenization, failing to accurately reflect diverse cultural values. This work proposes the first generalizable framework for cultural value assessment and intervention: leveraging the World Values Survey to construct contextualized moral dilemma probes, it maps a model’s internal cultural coordinates through implicit token probability analysis and implements activation steering during forward propagation to achieve cultural alignment without retraining. Experiments across multiple mainstream large language models demonstrate the method’s efficacy, reveal substantial variation in models’ cultural adaptability, and uncover—for the first time—an implicit coupling among cultural dimensions, termed “cultural entanglement,” which poses new challenges for precise cultural alignment.

Abstract

Large Language Models (LLMs) often exhibit homogenized cultural perspectives. While the World Values Survey (WVS) provides a gold standard for mapping human values, traditional direct prompting of LLMs on WVS often fails to access the model's latent cultural depth, leading to safety-aligned refusals or neutral responses. Here, we propose a generalizable framework for cultural evaluation and intervention that transitions from abstract queries to scenario-based behavioral probing. By extracting implicit token probabilities across 300 situational dilemmas, we bypass surface-level alignment to map the latent coordinates of LLMs cultural value. We further introduce activation steering to shift these internal alignments during the forward pass without retraining. Across multiple LLMs, we find substantial variation in adaptability and uncover a consistent phenomenon of latent entanglement, where interventions along one cultural dimension induce shifts along another. These results suggest that cultural values are encoded as coupled structures, limiting precise alignment. This work establishes a computationally efficient framework for cultural steering, highlighting the structural complexities when navigating global value with LLMs.

Problem

• Cultural Value Alignment
• Large Language Models
• World Values Survey
• Latent Representation
• Value Entanglement

Innovation

• activation steering
• cultural value alignment
• latent entanglement
• behavioral probing
• large language models