Break the Checkbox: Challenging Closed-Style Evaluations of Cultural Alignment in LLMs
A large number of studies rely on closed-style multiple-choice surveys to evaluate cultural alignment in Large Language Models (LLMs). In this work, we challenge this
