AI Paper Atlas

Research map / Large language models / AI and society

Quantifying the Salience of Geo-Cultural Values for Pluralistic Safety Alignment

Arkadiy Saakyan, Charvi Rastogi, Lora Aroyo

ICML 2026 · Large language models › AI and society · approach: Data commons and human baselines · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add cultural zone as fixed effect to base rater/item modelmultilevel safety rating model · 6 of 8 safety datasetsmodel fit (Delta AIC): Delta AIC -6.23 to -195.93Improved
add demographics (ethnicity, age, gender) as fixed effectsmultilevel safety rating model · 8 safety datasetsmodel fit (Delta AIC): Delta AIC -6.85 to -45.11Improved
add cultural zones on top of demographics (ablation)demographics-only multilevel model · 6 of 8 datasets (not DIVE, NLPos)Delta AIC: Delta AIC -3.97 to -179.88Improved
add culture x demographics interaction terms (ablation)D+CZ multilevel model · 7 of 8 datasetsDelta AIC / LRT: D3 only: -94.91, p < 0.001No clear effect
use culturally sensitive item triage with fine-tuned LMs (safe vs sensitive instead of safe vs unsafe)DeBERTa and Gemma-3-4B classifiers · D3F1: about 16% drop DeBERTa; about 14% drop GemmaHurt
scale from DeBERTa-Large (435M) to Gemma-3 4B decoder (ablation)quadrant rating emulation · DICES-990, CREHate, D3F1No clear effect
prompt reasoning LLMs (Gemini-3 Flash, GPT-5 Nano) to emulate cultural quadrantszero-shot LLM raters · D3F1 vs always-unsafe baselineNo clear effect
train on safe-vs-sensitive task and transfer to safe-vs-unsafe (ablation)fine-tuned classifiers · D3F1 cross-taskImproved
train classifier on D3 and test on other datasets (ablation)fine-tuned classifiers · CREHate, DICES-990F1No clear effect
raise Siq threshold from 0.5 to 0.7 (ablation)cultural sensitivity score · 6 datasets meanculturally sensitive item rate: 10.53 -> 3.43Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy