Learning to Judge: LLMs Designing and Applying Evaluation Rubrics

TL;DR

This paper explores whether large language models can autonomously create and apply their own evaluation rubrics for assessing language quality, revealing their strengths and limitations in different contexts.

Contribution

It introduces GER-Eval, a method for LLMs to design and use their own evaluation criteria, highlighting their capabilities and challenges in reliable assessment.

Findings

LLMs generate interpretable, task-aware evaluation criteria.

Scoring reliability drops in factual, knowledge-intensive tasks.

GPT-4o outperforms open models in agreement and generalization.

Abstract

Large language models (LLMs) are increasingly used as evaluators for natural language generation, applying human-defined rubrics to assess system outputs. However, human rubrics are often static and misaligned with how models internally represent language quality. We introduce GER-Eval (Generating Evaluation Rubrics for Evaluation) to investigate whether LLMs can design and apply their own evaluation rubrics. We evaluate the semantic coherence and scoring reliability of LLM-defined criteria and their alignment with human criteria. LLMs reliably generate interpretable and task-aware evaluation dimensions and apply them consistently within models, but their scoring reliability degrades in factual and knowledge-intensive settings. Closed-source models such as GPT-4o achieve higher agreement and cross-model generalization than open-weight models such as Llama. Our findings position evaluation as a learned linguistic capability of LLMs, consistent within models but fragmented across them, and call for new methods that jointly model human and LLM evaluative language to improve reliability and interpretability.