arXiv:2605.27016v1 Announce Type: cross 
Abstract: Large language models (LLMs) are prone to hallucinations, i.e., statements unsupported by the input or training data, hindering reliable deployment. In parallel, numerous uncertainty estimation (UE) methods have been proposed to quantify model confidence and are often implicitly treated as proxies for model failure. However, the relationship between uncertainty and hallucinations remains insufficiently characterized. We present a systematic empirical study of the association between uncertainty estimators and hallucinations in LLMs. Rather than assuming this association, we evaluate directly when and to what extent it holds. We consider a diverse set of uncertainty estimators, including information-theoretic, sampling-based, and reflexive estimators, and examine their behavior across hallucination settings. Our experiments cover both intrinsic hallucinations (violations of input faithfulness) and extrinsic hallucinations (unsupported claims relative to training data), using four complementary benchmarks, including RAGTruth and HalluLens. We find that the association is highly variable and often weak, depending on the hallucination type and the LLM under evaluation. These results challenge the use of uncertainty as a direct signal of hallucination and clarify when it provides actionable information.

تم إجراء دراسة تجريبية منهجية لتقييم العلاقة بين مقدرات عدم اليقين والهلاوس في نماذج اللغة الكبيرة (LLMs)، حيث تتناول التحديات التي تطرحها الهلاوس التي تعيق النشر الموثوق. تفحص الدراسة مجموعة متنوعة من طرق تقدير عدم اليقين، بما في ذلك الأساليب القائمة على نظرية المعلومات والأساليب القائمة على العينة، لفهم فعاليتها في التنبؤ بفشل النموذج.

Se ha llevado a cabo un estudio empírico sistemático para evaluar la relación entre los estimadores de incertidumbre y las alucinaciones en los grandes modelos de lenguaje (LLMs), abordando los desafíos que presentan las alucinaciones que obstaculizan un despliegue fiable. El estudio examina varios métodos de estimación de incertidumbre, incluidas las aproximaciones basadas en la teoría de la información y el muestreo, para comprender su efectividad en la predicción de fallos del modelo.

Une étude empirique systématique a été menée pour évaluer la relation entre les estimateurs d'incertitude et les hallucinations dans les grands modèles de langage (LLMs), abordant les défis posés par les hallucinations qui entravent un déploiement fiable. L'étude examine diverses méthodes d'estimation de l'incertitude, y compris des approches basées sur la théorie de l'information et l'échantillonnage, pour comprendre leur efficacité à prédire les échecs du modèle.

A systematic empirical study has been conducted to evaluate the relationship between uncertainty estimators and hallucinations in large language models (LLMs), addressing the challenges posed by hallucinations that hinder reliable deployment. The study examines various uncertainty estimation methods, including information-theoretic and sampling-based approaches, to understand their effectiveness in predicting model failures.