RubricBench reveals that AI-generated evaluation rubrics significantly underperform human standards.
As Large Language Model (LLM) alignment evolves from simple completions to complex, highly sophisticated generation, Reward Models are increasingly shifting toward rubric-guided evaluation to mitigate surface-level biases. However, the community lacks a unified benchmark to assess this evaluation paradigm, as existing benchmarks lack both the discriminative complexity and the ground-truth rubric annotations required for rigorous analysis. To bridge this gap, we introduce RubricBench, a curated benchmark with 1,147 pairwise comparisons specifically designed to assess the reliability of rubric-based evaluation. Our construction employs a multi-dimensional filtration pipeline to target hard samples featuring nuanced input complexity and misleading surface bias, augmenting each with expert-annotated, atomic rubrics derived strictly from instructions. Comprehensive experiments reveal a substantial capability gap between human-annotated and model-generated rubrics, indicating that even state-of-the-art models struggle to autonomously specify valid evaluation criteria, lagging considerably behind human-guided performance.
RubricBench is a benchmark with 1,147 preference pairs designed to assess rubric-guided evaluation reliability in reward models.
Model-generated rubrics lag 27% behind human-annotated rubrics, indicating current LLMs struggle with autonomous evaluation criteria specification.
Rubric-aware reward models reach 58% accuracy versus 40-47% for previous approaches, validating the rubric-guided evaluation paradigm.
Benchmark is limited to 1,147 samples, which may not fully represent all complex reasoning-intensive tasks LLMs encounter.
Study focuses on rubric generation but doesn't explore how to systematically improve model ability to generate human-aligned rubrics.
Chenxu Wang, Chaozhuo Li +11
The emergence of multi-agent systems built from large language models (LLMs) offers a promising paradigm for scalable collective intelligence and self-evolution. Ideally, such systems would achieve co...
Dongrui Liu, Qihan Ren +41
The rise of AI agents introduces complex safety and security challenges arising from autonomous tool use and environmental interactions. Current guardrail models lack agentic risk awareness and transp...
Seanie Lee, Sangwoo Park +7
Large reasoning models (LRMs) achieve remarkable performance by leveraging reinforcement learning (RL) on reasoning tasks to generate long chain-of-thought (CoT) reasoning. However, this over-optimiza...
Dongrui Liu, Yi Yu +19
To understand and identify the unprecedented risks posed by rapidly advancing artificial intelligence (AI) models, Frontier AI Risk Management Framework in Practice presents a comprehensive assessment...
Rakshith Vasudev, Melisa Russak +2
Proactive interventions by LLM critic models are often assumed to improve reliability, yet their effects at deployment time are poorly understood. We show that a binary LLM critic with strong offline ...
