A Survey of Uncertainty Estimation Methods on Large Language Models

2025-02-28

AI Summary

Large language models (LLMs) exhibit strong generative capabilities but frequently produce outputs containing bias, hallucinations, and factual inaccuracies—highlighting the critical need for reliable uncertainty estimation. To address this gap, we present the first systematic survey of uncertainty estimation for LLMs, proposing a unified four-dimensional taxonomy: sampling-based, logits/probability-based, reasoning-process-based, and external-calibration-based methods. We establish a reproducible benchmarking protocol and conduct extensive empirical evaluation across 12 state-of-the-art LLMs and 6 diverse datasets. Our results delineate the effectiveness boundaries of each method class in detecting hallucinations, biases, and factual errors. This work fills a key void in the literature by providing the first comprehensive survey of LLM uncertainty estimation, offering both a methodological foundation for trustworthy AI and concrete directions for future research.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Uncertainty Representations
• Natural Language Processing: (Large) Language Models

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information

Abstract

Large language models (LLMs) have demonstrated remarkable capabilities across various tasks. However, these models could offer biased, hallucinated, or non-factual responses camouflaged by their fluency and realistic appearance. Uncertainty estimation is the key method to address this challenge. While research efforts in uncertainty estimation are ramping up, there is a lack of comprehensive and dedicated surveys on LLM uncertainty estimation. This survey presents four major avenues of LLM uncertainty estimation. Furthermore, we perform extensive experimental evaluations across multiple methods and datasets. At last, we provide critical and promising future directions for LLM uncertainty estimation.

Problem

• Address biased and non-factual responses in LLMs
• Survey methods for uncertainty estimation in LLMs
• Evaluate and propose future directions for LLM uncertainty

Innovation

• Survey of uncertainty estimation methods for LLMs
• Experimental evaluations across multiple methods
• Future directions for LLM uncertainty estimation