analiza aiqbench
This research introduces C-Voices, a novel multilingual dataset designed to probe how Large Language Models handle culturally specific Chinese social values across six languages. The authors demonstrate that these value preferences are inherently language-sensitive and model-dependent, then propose a fine-tuning-free steering technique using hidden-state discrepancies to align outputs with desired cultural norms. This work advances the critical field of AI alignment by addressing the complexity of pluralistic, culturally grounded values beyond standard safety constraints.
Same Values, Different Languages? From Multilingual Probing to Steering LLMs Toward Chinese Social Values
arXiv:2609.08515v1 Announce Type: new Abstract: As Large Language Models (LLMs) are increasingly integrated into human society, aligning them with pluralistic social values has become a critical priority. However, whether LLMs exhibit consistent value preferences across languages remains underexplored, particularly for culturally grounded values, which are more abstract and difficult to evaluate and align than safety-centric principles. We investigate this issue through Chinese Social Values (CSV), a value system rooted in Chinese culture and comprising $12$ dimensions across national, societal, and personal levels. We construct C-Voices, the first comprehensive multilingual contrastive probe dataset for CSV, with 86,400 dilemma-based instances in six languages, each pairing a CSV-aligned action with a value-conflicting alternative. Building on the contrastive probes of C-Voices, we then propose a fine-tuning-free value vector steering method that derives value directions from hidden-state discrepancies and selectively intervenes on value-sensitive layers during inference. Experiments on six languages show that CSV-oriented preferences are model-dependent and language-sensitive, with the same dilemma eliciting divergent responses across languages. Our method achieves effective CSV steering, supports cross-lingual transfer of value vectors, and generalizes to existing FLAMES and ValuePrism.
Title:Same Values, Different Languages? From Multilingual Probing to Steering LLMs Toward Chinese Social Values
Abstract:As Large Language Models (LLMs) are increasingly integrated into human society, aligning them with pluralistic social values has become a critical priority. However, whether LLMs exhibit consistent value preferences across languages remains underexplored, particularly for culturally grounded values, which are more abstract and difficult to evaluate and align than safety-centric principles. We investigate this issue through Chinese Social Values (CSV), a value system rooted in Chinese culture and comprising $12$ dimensions across national, societal, and personal levels. We construct C-Voices, the first comprehensive multilingual contrastive probe dataset for CSV, with 86,400 dilemma-based instances in six languages, each pairing a CSV-aligned action with a value-conflicting alternative. Building on the contrastive probes of C-Voices, we then propose a fine-tuning-free value vector steering method that derives value directions from hidden-state discrepancies and selectively intervenes on value-sensitive layers during inference. Experiments on six languages show that CSV-oriented preferences are model-dependent and language-sensitive, with the same dilemma eliciting divergent responses across languages. Our method achieves effective CSV steering, supports cross-lingual transfer of value vectors, and generalizes to existing FLAMES and ValuePrism.
Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
arXiv:2609.08515 [cs.CL]
(or arXiv:2609.08515v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2609.08515 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 288 cuvinte
Subiecte conexe
J++ Lens: Jacobian Filtering Enables More Faithful Workspace Lenses
LessWrong · 8 oct. 2026
PathCal: State-Aware Reflection-Marker Calibration for Efficient Reasoning
arXiv · 17 oct. 2026
Lingua Franca or Probing Artifact? Rethinking Latent Language in Multilingual LLMs
arXiv · 2 sept. 2026
Some Intuitions on Steering Vectors
LessWrong · 29 sept. 2026