Previous
Human-Centered Evaluation of LLM-Generated Code Quality
Human evaluation of LLM-generated code is often inconsistent across subjective quality dimensions such as readability, modularity, style, understandability, and effort. Ambiguous evaluation guidelines can reduce reproducibility and make benchmark results difficult to compare.
Project Summary
This work studied rubric-guided, AI-assisted human evaluation across multiple programming languages. The research focused on reducing ambiguity-driven disagreement while preserving human judgment, improving the reproducibility and structure of code-quality assessments.