Agentic Uncertainty Quantification

2026-01-22 · Jiaxin Zhang, Prafulla Kumar Choubey, Kung-Hsiang Huang, Caiming Xiong, Chien-Sheng Wu arxiv

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.

📄 PDF Abstract BibTeX arXiv:2601.15703

Code (0)

등록된 구현이 없습니다.

Uncertainty Quantification in LLM Agents: Foundations, Emerging Challenges, and Opportunities

2026-02-04 · Changdae Oh, Seongheon Park, To Eun Kim, Jiatong Li 외 arxiv

Uncertainty quantification (UQ) for large language models (LLMs) is a key building block for safety guardrails of daily LLM applications. Yet, even as LLM agents are increasingly deployed in highly complex tasks, most UQ…

Directional Concentration Uncertainty: A representational approach to uncertainty quantification for generative models

2026-02-04 · Souradeep Chattopadhyay, Brendan Kennedy, Sai Munikoti, Soumik Sarkar 외 arxiv

In the critical task of making generative models trustworthy and robust, methods for Uncertainty Quantification (UQ) have begun to show encouraging potential. However, many of these methods rely on rigid heuristics that …

Seeing and Reasoning with Confidence: Supercharging Multimodal LLMs with an Uncertainty-Aware Agentic Framework

2025-03-11 · Zhuo Zhi, Chen Feng, Adam Daneshmend, Mine Orlu 외

Multimodal large language models (MLLMs) show promise in tasks like visual question answering (VQA) but still face challenges in multimodal reasoning. Recent works adapt agentic frameworks or chain-of-thought (CoT) reaso…

Conformal Prediction Multimodal Reasoning Question Answering Uncertainty Quantification+2

UProp: Investigating the Uncertainty Propagation of LLMs in Multi-Step Agentic Decision-Making

2025-06-20 · Jinhao Duan, James Diffenderfer, Sandeep Madireddy, Tianlong Chen 외

As Large Language Models (LLMs) are integrated into safety-critical applications involving sequential decision-making in the real world, it is essential to know when to trust LLM decisions. Existing LLM Uncertainty Quant…

Decision Making Question Answering Sequential Decision Making Uncertainty Quantification

Exploring Agentic Tool-Calling Decisions via Uncertainty-Aligned Reinforcement Learning

2026-06-05 · Yijin Zhou, Linqian Zeng, Xiaoya Lu, Wenyuan Xie 외 arxiv

Large language model (LLM)-based agents often make suboptimal tool-use decisions, including unsupported tool invocation and hallucinated direct responses, which may accumulate errors throughout multi-step interactions. E…

Reinforcement Learning