Do RAG Systems Really Suffer From Positional Bias?
This work addresses the problem of positional bias in RAG systems for AI researchers and practitioners, showing it is incremental by revealing that existing concerns are overstated due to retrieval issues.
This paper investigates positional bias in Retrieval Augmented Generation (RAG) systems, finding that in real scenarios, the bias has marginal impact because retrieval pipelines often bring distracting passages to top ranks, which are penalized along with relevant ones, with over 60% of queries containing highly distracting passages in the top-10 retrieved.
Retrieval Augmented Generation enhances LLM accuracy by adding passages retrieved from an external corpus to the LLM prompt. This paper investigates how positional bias - the tendency of LLMs to weight information differently based on its position in the prompt - affects not only the LLM's capability to capitalize on relevant passages, but also its susceptibility to distracting passages. Through extensive experiments on three benchmarks, we show how state-of-the-art retrieval pipelines, while attempting to retrieve relevant passages, systematically bring highly distracting ones to the top ranks, with over 60% of queries containing at least one highly distracting passage among the top-10 retrieved passages. As a result, the impact of the LLM positional bias, which in controlled settings is often reported as very prominent by related works, is actually marginal in real scenarios since both relevant and distracting passages are, in turn, penalized. Indeed, our findings reveal that sophisticated strategies that attempt to rearrange the passages based on LLM positional preferences do not perform better than random shuffling.