Risk Management for Mitigating Benchmark Failure Modes: BenchRisk

Sean McGregor, Vassil Tashev, Armstrong Foundjem, Aishwarya Ramasethu, Sadegh AlMahdi Kazemi Zarkouei, Chris Knotz, Kongtao Chen, Alicia Parrish, Anka Reuel, Heather Frase

Keywords: Risk Management, Risk Analysis, Risk Mitigation, Human Factors, Reliability Engineering, Benchmarks, Metaevaluation, Deployment, Safety, Assessment

TL;DR: Risk management processes as a way of improving, assessing, and comparing benchmark reliability result in a benchmark of benchmarks

Abstract: Large language model (LLM) benchmarks inform LLM use decisions (e.g., "is this LLM safe to deploy for my use case and context?"). However, benchmarks may be rendered unreliable by various failure modes impacting benchmark bias, variance, coverage, or people's capacity to understand benchmark evidence. Using the National Institute of Standards and Technology's risk management process as a foundation, this research iteratively analyzed 26 popular benchmarks, identifying 57 potential failure modes and 196 corresponding mitigation strategies. The mitigations reduce failure likelihood and/or severity, providing a frame for evaluating "benchmark risk," which is scored to provide a metaevaluation benchmark: BenchRisk. Higher scores indicate benchmark users are less likely to reach an incorrect or unsupported conclusion about an LLM. All 26 scored benchmarks present significant risk within one or more of the five scored dimensions (comprehensiveness, intelligibility, consistency, correctness, and longevity), which points to important open research directions for the field of LLM benchmarking. The BenchRisk workflow allows for comparison between benchmarks; as an open-source tool, it also facilitates the identification and sharing of risks and their mitigations.

Code URL: https://github.com/BenchRisk/BenchRisk

Primary Area: Social and economic aspects of datasets and benchmarks in machine learning (e.g., fairness, interpretability, human-AI interaction, privacy, safety, strategic behavior)

Submission Number: 1411