From Inertia to Objectivity: Improving Deep Research Agents with Noise Isolation
2026-08-24 • Artificial Intelligence
Artificial Intelligence
AI summaryⓘ
The authors studied a problem called inertia bias, where AI agents become less objective when judging actions they themselves took during web searches. They created the IBIS benchmark to measure this bias by controlling search results and seeing how models assess their own previous steps. Their findings show that self-judgment leads to worse decisions and propagates errors at different system levels. To fix this, they developed NIS-Agent, which isolates context during key decision points, lowering costs and improving performance on various search tasks. They also trained a model less prone to this bias, achieving results close to GPT-4o on complex research challenges.
Large Language ModelsWeb Search AgentsInertia BiasIBIS benchmarkContext IsolationWebpage TriageFinal-Answer ValidationSearch NoiseSystem-level DegradationGPT-4o
Authors
Xiangxin Zhang, Zhanwei Zhang, Zhihang Fu, Binbin Lin, Wenxiao Wang
Abstract
Web search agents powered by Large Language Models (LLMs) show strong promise, but deep research tasks expose a recurring failure mode: once an agent has produced a query, plan, or intermediate conclusion, it becomes less objective when later judging the consequences of that same action. We term this phenomenon \textbf{inertia bias}. To make it measurable, we introduce the IBIS benchmark, which controls the search observations while varying whether the model is evaluating the outcome of its own prior action. We find that models are substantially worse when they ``own'' the preceding search step, showing that self-authored action history can systematically distort subsequent judgment. We further show that this bias propagates into two forms of system-level degradation: search noise at the worker level and contextual noise at the manager level. To address this problem, we propose NIS-Agent, which applies context isolation at the two decision points most vulnerable to inertia bias: webpage triage and final-answer validation. Across GAIA, WebWalkerQA, BrowseComp, and BrowseComp-zh, NIS-Agent achieves competitive performance while reducing token cost by 33\% compared to our baseline. We further train an 8B model to be intrinsically more resistant to inertia bias; under the same NIS-Agent framework, it attains average performance comparable to GPT-4o on deep research benchmarks.