Agentic Uncertainty Quantification

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Zhang, Jiaxin, Choubey, Prafulla Kumar, Huang, Kung-Hsiang, Xiong, Caiming, Wu, Chien-Sheng
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866911391923830784
author Zhang, Jiaxin
Choubey, Prafulla Kumar
Huang, Kung-Hsiang
Xiong, Caiming
Wu, Chien-Sheng
author_facet Zhang, Jiaxin
Choubey, Prafulla Kumar
Huang, Kung-Hsiang
Xiong, Caiming
Wu, Chien-Sheng
contents Although AI agents have demonstrated impressive capabilities in long-horizon reasoning, their reliability is severely hampered by the ``Spiral of Hallucination,'' where early epistemic errors propagate irreversibly. Existing methods face a dilemma: uncertainty quantification (UQ) methods typically act as passive sensors, only diagnosing risks without addressing them, while self-reflection mechanisms suffer from continuous or aimless corrections. To bridge this gap, we propose a unified Dual-Process Agentic UQ (AUQ) framework that transforms verbalized uncertainty into active, bi-directional control signals. Our architecture comprises two complementary mechanisms: System 1 (Uncertainty-Aware Memory, UAM), which implicitly propagates verbalized confidence and semantic explanations to prevent blind decision-making; and System 2 (Uncertainty-Aware Reflection, UAR), which utilizes these explanations as rational cues to trigger targeted inference-time resolution only when necessary. This enables the agent to balance efficient execution and deep deliberation dynamically. Extensive experiments on closed-loop benchmarks and open-ended deep research tasks demonstrate that our training-free approach achieves superior performance and trajectory-level calibration. We believe this principled framework AUQ represents a significant step towards reliable agents.
format Preprint
id arxiv_https___arxiv_org_abs_2601_15703
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Agentic Uncertainty Quantification
Zhang, Jiaxin
Choubey, Prafulla Kumar
Huang, Kung-Hsiang
Xiong, Caiming
Wu, Chien-Sheng
Artificial Intelligence
Computation and Language
Although AI agents have demonstrated impressive capabilities in long-horizon reasoning, their reliability is severely hampered by the ``Spiral of Hallucination,'' where early epistemic errors propagate irreversibly. Existing methods face a dilemma: uncertainty quantification (UQ) methods typically act as passive sensors, only diagnosing risks without addressing them, while self-reflection mechanisms suffer from continuous or aimless corrections. To bridge this gap, we propose a unified Dual-Process Agentic UQ (AUQ) framework that transforms verbalized uncertainty into active, bi-directional control signals. Our architecture comprises two complementary mechanisms: System 1 (Uncertainty-Aware Memory, UAM), which implicitly propagates verbalized confidence and semantic explanations to prevent blind decision-making; and System 2 (Uncertainty-Aware Reflection, UAR), which utilizes these explanations as rational cues to trigger targeted inference-time resolution only when necessary. This enables the agent to balance efficient execution and deep deliberation dynamically. Extensive experiments on closed-loop benchmarks and open-ended deep research tasks demonstrate that our training-free approach achieves superior performance and trajectory-level calibration. We believe this principled framework AUQ represents a significant step towards reliable agents.
title Agentic Uncertainty Quantification
topic Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2601.15703