Study: LLMs show clear cultural bias toward Japan, not just the West
JeremyNguyenPhD · x · 2026-10-30
A study from the University of the Basque Country and Cardiff University ("Why are all LLMs Obsessed with Japanese Culture?") challenges the assumption that AI is purely Western-biased. Using the CROQ dataset of open culture questions, the authors find LLMs display a clear tendency toward countries like Japan. Higher-resource prompt languages yield more diverse answers, and the bias emerges after supervised fine-tuning rather than during pre-training.
More from Research
- IBM/Cornell's LPT shares KV-cache across looped transformer passes, cutting memory 79% while improving quality — burny_tech · 2026-10-30
- Artificial skin lets robot fingers grasp pencil lead and dodge needles, reading 10,000+ sensors 100x/sec — CurieuxExplorer · 2026-10-30
- Theorem Provers vs Specification Languages: How to Pick Between Lean, TLA+, P and More — burny_tech · 2026-10-30
- Integer factorization proven subexponential, with core work done by Claude — burny_tech · 2026-10-30
- Adversarial Blinding trains away a model's ability to detect evals — burny_tech · 2026-10-30
- Paper Argues One-Click AI Video Will Break Recommendation Pools, Proposes BoLe System — BreadStudio · 2026-10-30