Comparing Human and Large Language Model Performance on Visual Analogies

LLMs struggle with visual analogies compared to humans, showing different error patterns.