Tool Use

Ask about this topic

Overview

Tool use in AI studies how language and multimodal agents choose tools such as APIs, code interpreters, search systems, and robotic actions, then use their results to complete tasks. Common approaches train agents on synthetic or verified interaction traces, use reinforcement learning to improve multi-step planning and tool selection, and add runtime mechanisms for state tracking, verification, and recovery. Benchmarks test function calling, long workflows, multimodal and embodied tasks, and performance when tools or environments are unreliable. A central challenge is making tool use both selective and dependable: agents must know when a tool helps, execute calls correctly, and recover from failures without unnecessary calls or harmful side effects.

Latest papers

CardsList

91 papers11 papers in the last four weeks, level with the four weeks before. 0.1% of all new papers.

Aug 21, 2025

When Tools Hurt LLM Reasoning: State-Dependent Belief Revision under External Evidence

Tool use is often assumed to monotonically improve reasoning, where external evidence is expected to help when relevant and be ignored when irrelevant. We show that this assumption fails in a state-dependent way. Across benchmarks with Python and Wikipedia tools, external evidence reliably helps when initial beliefs are weak, but can flip already-correct answers when those beliefs are strong. We frame this as a misallocation of revision authority, arguing that deferring to external evidence is suboptimal when internal support for the correct answer surpasses the tool's expected output quality. This predicts that harm should concentrate on high-confidence no-tool cases. We test this prediction with threshold localization, wrong-trace audits, and a same-clue intervention showing that revision framing changes the damage caused by misleading evidence. These findings suggest that mixed no-tool/tool-assisted inference should arbitrate authority rather than privilege tool evidence by default. As a minimal demonstration, we introduce CASE, a label-free controller that selects between no-tool and tool-assisted trajectories using answer-state certainty and improves over existing confidence-based arbitration baselines.The code for our experiments is available at https://github.com/epsilondylan/State-Dependent-Belief-Revision.

Yufeng Zhao, Junnan Liu, Hongwei Liu +4

Shanghai Artificial Intelligence Laboratory · Department of Electronic Engineering, Tsinghua University · Beijing National Research Center for Information Science and Technology +2

Belief RevisionKnowledge Conflicts in Language ModelsLLM Decision-MakingLLM Tool UseTool Use
AskPDF