The paper introduces a framework for assessing how well language models (LMs) calibrate their confidence when answering relational knowledge questions. It evaluates three types of confidence—intrinsic, paraphrase consistency, and linguistic—across 21 causal and 8 masked LMs, finding that most models, especially masked ones, are overconfident. The study shows that confidence estimates that account for rephrasing inconsistencies and increasing answer options lead to better calibration.
By Christopher Kissling, Elena Merdjanovska, Alan Akbik
As large language models (LLMs) grow more capable, they are increasingly deployed in context-rich settings where task inputs are often accompanied by long, partially irrelevant context. In a controlled setting, we find that state-of-the-art models often appear robust to task-irrelevant context at the aggregate level: prepending it to benchmark questions causes little change in overall accuracy.
The study examines how the large language model Qwen3-4B exhibits overconfidence across three expressions of uncertainty—verbal epistemic markers, abstention, and numeric confidence scores—using controlled reasoning scenarios that vary logical necessity and possibility. Findings confirm a pronounced overconfidence tendency, especially when numeric scores are requested, and reveal that certainty is generated by a broad coalition of shared features while uncertainty relies on a sparse set of dedicated features. By intervening on these uncertainty features, the researchers demonstrate causal evidence of the imbalance driving overconfidence and show that this intervention can mitigate overconfident errors across different settings, languages, and an out‑of‑distribution modality task.
By Davide Mazzaccara, Leonardo Bertolazzi, Raffaella Bernardi
Large language models (LLMs) frequently produce confident yet factually incorrect responses when user inputs contain misleading premises, a phenomenon we attribute to fact perturbations in the input....
Large language models are informing decisions with ever-higher stakes. As the consequences of their errors grow, a central question becomes harder to ignore: how much can we trust an individual answer...
arXiv:2604.08974v2 Announce Type: replace
Abstract: Uncertainty quantification techniques measure confidence in language model outputs to support critical applications like hallucination detection an...
By Lorenzo Jaime Yu Flores, Cesare Spinoso di-Piano, Jackie Chi Kit Cheung
arXiv:2606. 02093v1 Announce Type: cross Abstract: The task of Error Prediction, namely predicting whether a model output is correct, is commonly tackled with Uncertainty Quantification (UQ).
By Ieva Raminta Stali\=unait\.e, James Bishop, Andreas Vlachos
The paper investigates the relationship between a large language model’s internal probability distribution and its verbalized confidence statements. By systematically manipulating training and in‑context data, the authors show that both internal and verbalized probabilities are influenced by distributional and asserted uncertainty in the data. They find that verbalized probabilities align with internal ones beyond what would be expected if they tracked the same sources independently, indicating that verbalized confidence can serve as a probe of the model’s internal distribution.
By Sinead Williamson, Jiaxuan Li, Nick Foti, Russ Webb, Masha Fedzechkina
The paper introduces DEDUCE, a three‑stage framework that turns large language models into proactive error correctors by detecting input fact errors, devising correction strategies, and delivering reliable answers. It also presents MisFactQA, a dataset of factual errors, and new metrics for robustness evaluation. Experiments on TruthfulQA, FalseQA, and MisFactQA show significant gains in accuracy and error correction across Qwen, LLaMA, and Gemma models.
By Ping Wang, Xiangguo Sun, Bingbing Xu, Guocong Li, Xiaofeng Meng
Recent advancements in Large Language Models (LLMs) have enabled sophisticated reasoning and content generation, yet their inherent stochasticity poses significant challenges for ensuring predictive credibility. While traditional uncertainty taxonomy paradigms, such as the dichotomy of aleatoric and epistemic uncertainties, provide conceptual foundations, they often fail to capture the multi-component and multi-stage nature of LLM generation and struggle to evaluate the effectiveness of various Uncertainty Quantification (UQ) methods.
arXiv:2607. 19367v1 Announce Type: new Abstract: Calibration is the primary criterion for evaluating LLM confidence, but it is insufficient: it admits trivially incoherent estimators, depends on the evaluation distribution, and does not test the extent to which the estimation can be interpreted as a consistent, underlying probability function.
By Krish Matta, Atharv Naphade, Andy Zou
arXiv:2609.24238v1 Announce Type: new
Abstract: We reproduce and stress-test the work of Yu et al. (2023), who characterize how language models (LMs) arbitrate between memorized knowledge and contrad...
By Guilhem Fouilh\'e, Nicholas Asher, Philippe Muller