PredictaBoard: Benchmarking LLM Score Predictability

2025-02-20

AI Summary

Large language models (LLMs) exhibit high variance in error rates on commonsense reasoning tasks, undermining their safe deployment. Method: This paper introduces PredictaBoard, the first benchmark framework centered on *instance-level error predictability*—i.e., predicting the probability that a model will err on a given prompt—to enable “foreseeable safety.” It establishes an instance-level evaluation protocol and quantifies predictor performance via rejection-rate–error-tolerance curves, supporting multi-model and multi-evaluator assessment. Contribution/Results: Experiments reveal that existing baseline predictors perform poorly, underscoring the critical need for predictive error assessment. PredictaBoard provides a novel, open-source evaluation paradigm for trustworthy, human-intervenable AI systems, shifting LLM safety evaluation from aggregate performance metrics toward fine-grained, controllable error prediction.

Technology Category

• Natural Language Processing: Safety and Robustness
• Machine Learning: Large Multimodal Models (LMMs)
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Despite possessing impressive skills, Large Language Models (LLMs) often fail unpredictably, demonstrating inconsistent success in even basic common sense reasoning tasks. This unpredictability poses a significant challenge to ensuring their safe deployment, as identifying and operating within a reliable"safe zone"is essential for mitigating risks. To address this, we present PredictaBoard, a novel collaborative benchmarking framework designed to evaluate the ability of score predictors (referred to as assessors) to anticipate LLM errors on specific task instances (i.e., prompts) from existing datasets. PredictaBoard evaluates pairs of LLMs and assessors by considering the rejection rate at different tolerance errors. As such, PredictaBoard stimulates research into developing better assessors and making LLMs more predictable, not only with a higher average performance. We conduct illustrative experiments using baseline assessors and state-of-the-art LLMs. PredictaBoard highlights the critical need to evaluate predictability alongside performance, paving the way for safer AI systems where errors are not only minimised but also anticipated and effectively mitigated. Code for our benchmark can be found at https://github.com/Kinds-of-Intelligence-CFI/PredictaBoard

Problem

• Evaluates LLM error predictability
• Develops safer AI deployment strategies
• Benchmarks assessors for anticipating LLM failures

Innovation

• Collaborative benchmarking framework
• Evaluates LLM error predictability
• Uses rejection rate metrics