How LLMs See Creativity: Zero-Shot Scoring of Visual Creativity with Interpretable Reasoning

2026

Open publication workspace · Sign in to read the full PDF.

AI-generated summary

1. This research demonstrates that large language models (LLMs) can accurately assess visual creativity in a zero-shot manner, providing interpretable reasoning for their judgments.
2. The study evaluates six multimodal LLMs on their ability to score AI-generated images and hand-drawn sketches for creativity, analyzing their reasoning processes to understand how they arrive at their ratings.
3. Findings show LLMs can match human creativity judgments without fine-tuning, and their reasoning chains offer insights into their evaluative criteria, though reasoning itself doesn't improve accuracy.
Tags: LLM-as-a-judge, visual creativity, zero-shot scoring, interpretable AI, creativity assessment

Check the original publication for accuracy and context.