Can Unconfident LLM Annotations Be Used for Confident Conclusions?

2024-08-27

arXiv.org

AI Summary

This study addresses the problem of unreliable downstream statistical inference in computational social science caused by unstable annotation quality from large language models (LLMs). We propose Confidence-Driven Inference (CDI), a novel framework that jointly leverages LLM-generated annotations, LLM-assigned confidence scores, statistical learning theory, and active sampling. CDI constitutes the first hybrid annotation framework with formal theoretical guarantees: it ensures that statistical estimation accuracy and confidence interval coverage are no worse than those achieved under purely human annotation, while intelligently selecting only low-confidence samples for human verification. Evaluated on three canonical tasks—politeness detection, stance classification, and bias identification—CDI reduces human annotation effort by over 25% while maintaining estimation accuracy and confidence interval coverage at levels satisfying statistical validity criteria.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Other Foundations of Reasoning under Uncertainty
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning

Application Category

• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Web Mining and Content Analysis: Web data provenance, reliability, and authenticity

Abstract

Large language models (LLMs) have shown high agreement with human raters across a variety of tasks, demonstrating potential to ease the challenges of human data collection. In computational social science (CSS), researchers are increasingly leveraging LLM annotations to complement slow and expensive human annotations. Still, guidelines for collecting and using LLM annotations, without compromising the validity of downstream conclusions, remain limited. We introduce Confidence-Driven Inference: a method that combines LLM annotations and LLM confidence indicators to strategically select which human annotations should be collected, with the goal of producing accurate statistical estimates and provably valid confidence intervals while reducing the number of human annotations needed. Our approach comes with safeguards against LLM annotations of poor quality, guaranteeing that the conclusions will be both valid and no less accurate than if we only relied on human annotations. We demonstrate the effectiveness of Confidence-Driven Inference over baselines in statistical estimation tasks across three CSS settings--text politeness, stance, and bias--reducing the needed number of human annotations by over 25% in each. Although we use CSS settings for demonstration, Confidence-Driven Inference can be used to estimate most standard quantities across a broad range of NLP problems.

Problem

• Ensure validity of LLM annotations
• Reduce human annotation workload
• Improve statistical estimation accuracy

Innovation

• Combines LLM annotations with confidence indicators
• Reduces human annotations by over 25%
• Ensures valid and accurate statistical estimates