1 paper · 1 filter
Haorui Yu, Yang Zhao, Yijia Chu +1
Vision-Language Models (VLMs) often appear culturally competent but rely on superficial pattern matching rather than genuine cultural understanding. We introduce a diagnostic frame…