Astute RAG: Overcoming Imperfect Retrieval Augmentation and Knowledge Conflicts for Large Language Models

TL;DR

This work investigates the vulnerability of retrieval-augmented generation to imperfect retrieval and internal-external knowledge conflicts. It introduces Astute RAG, a training-free, black-box-friendly approach that adaptively elicits LLM internal knowledge, consolidates it with externally retrieved passages in a source-aware, iterative manner, and finalizes answers by assessing information reliability. Through extensive experiments on multiple LLMs and diverse datasets, Astute RAG consistently outperforms robustness-focused baselines, including under worst-case noisy retrieval where it approaches non-RAG performance. The analysis reveals that knowledge conflicts are a key bottleneck and that adaptive internal knowledge generation and source-aware consolidation effectively resolve these conflicts, enhancing both accuracy and trustworthiness. The method offers practical gains in robustness with minimal inference cost, highlighting its potential for real-world, knowledge-intensive applications.

Abstract

Retrieval augmented generation (RAG), while effectively integrating external knowledge to address the inherent limitations of large language models (LLMs), can be hindered by imperfect retrieval that contain irrelevant, misleading, or even malicious information. Previous studies have rarely connected the behavior of RAG through joint analysis, particularly regarding error propagation coming from imperfect retrieval and potential conflicts between LLMs' internal knowledge and external sources. Through comprehensive and controlled analyses under realistic conditions, we find that imperfect retrieval augmentation is inevitable, common, and harmful. We identify the knowledge conflicts between LLM-internal and external knowledge from retrieval as a bottleneck to overcome imperfect retrieval in the post-retrieval stage of RAG. To address this, we propose Astute RAG, a novel RAG approach designed to be resilient to imperfect retrieval augmentation. It adaptively elicits essential information from LLMs' internal knowledge, iteratively consolidates internal and external knowledge with source-awareness, and finalizes the answer according to information reliability. Our experiments with Gemini and Claude demonstrate the superior performance of Astute RAG compared to previous robustness-enhanced RAG approaches. Specifically, Astute RAG is the only RAG method that achieves performance comparable to or even surpassing conventional use of LLMs under the worst-case scenario. Further analysis reveals the effectiveness of Astute RAG in resolving knowledge conflicts, thereby improving the trustworthiness of RAG.

Figures (12)

• Figure 1: Knowledge conflicts between the LLMs' internal knowledge and retrieved knowledge from external sources. We report the overall results with Claude under the setting in Sec. \ref{['sec:experiment_setting']}.
• Figure 2: Imperfect retrieval (samples with low retrieval precision) is prevalent in real-world RAG.
• Figure 3: Conflicting rate between answers from LLMs with and without RAG on different retrieval precision.
• Figure 4: Overview of the Astute RAG framework. Astute RAG is designed to better combine the information from the external sources (e.g. web, domain-specific corpora) and internal knowledge of the LLMs by employing a consolidation mechanism to address the conflicts, which eventually leads to superior generation quality.
• Figure 5: Performance on ASQA.