Which Metrics Save the Most Human Annotation? Prediction-Powered Evaluation and Meta-Evaluation

@misc{gao2026which,
  title = {{Which Metrics Save the Most Human Annotation? Prediction-Powered Evaluation and Meta-Evaluation}},
  author = {Mingqi Gao and Anthony Sicilia and Weiyan Shi},
  howpublished = {arXiv},
  year = {2026},
  month = aug,
  doi = {10.48550/arxiv.2608.26638},
  eprint = {2608.26638},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2608.26638},
}

TY  - UNPB
TI  - Which Metrics Save the Most Human Annotation? Prediction-Powered Evaluation and Meta-Evaluation
AU  - Mingqi Gao
AU  - Anthony Sicilia
AU  - Weiyan Shi
PY  - 2026
DA  - 2026/08/30
PB  - arXiv
DO  - 10.48550/arxiv.2608.26638
UR  - https://arxiv.org/abs/2608.26638
AB  - Across various non-verifiable tasks, human evaluation is reliable but expensive, while automatic metrics are more scalable but often biased. Building on prediction-powered inference (PPI), we propose prediction-powered evaluation, a framework that combines limited human judgments with large-scale automatic scores to obtain data-efficient system comparisons that are provably unbiased. We develop parametric and non-parametric procedures, analyze the efficiency trade-off between paired and unpaired designs, and validate the framework on six WMT datasets. We further introduce the Prediction-Powered Saving Ratio (PPSR), a meta-metric that measures how much human annotation an automatic metric can save when used within prediction-powered evaluation. PPSR directly targets metric utility for prediction-powered evaluation and yields more discriminative and stable metric rankings than existing system-level meta-metrics. Overall, our new paradigm reframes automatic metrics as tools for reducing human annotation cost rather than replacing human judgment, and applies broadly to non-verifiable tasks.
ER  -