- Rating
- 3
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 3(+1)
Read review
Paper summary
This paper proposes a causal attribution framework for identifying where demographic disparities arise in RAG systems. It decomposes disparities across three components: historical data, retrieval, and LLM aggregation. It further attributes each contribution to direct, indirect, and spurious causal pathways, yielding a 3*3 bias decomposition. Experiments on semi-synthetic data validate the estimation accuracy, while evaluations on emergency triage (MIMIC-IV-ED) and microfinance lending (Kiva) show that retrieval and LLM aggregation can amplify, mitigate, or introduce disparities through different causal pathways, with the effects of LLM aggregation depending on the retrieved evidence.
Critical strengths
-
The paper provides a useful framework for diagnosing demographic disparities at both the component and causal pathway levels in RAG systems. By distinguishing the contributions of historical data, retrieval, and LLM aggregation, it offers a fine-grained and actionable insights than conventional fairness evaluations that focus on aggregate disparities.
-
The paper provides meaningful empirical validation through both semi-synthetic and real-world experiments. The semi-synthetic experiments demonstrate that the proposed estimators can recover known causal effects, while experiments on emergency triage and microfinance lending reveal how retrieval and LLM aggregation influence different causal pathways.
Critical weaknesses
- The proposed disparity attribution depends on the choice of the intermediate retrieval predictor. Different reasonable choices may lead to different component-level attributions, even when the final RAG outputs remain unchanged, raising concerns about the robustness and reliability of the attribution results.
Questions for authors
- How sensitive are the component-level attribution results to the choice of the intermediate retrieval predictor?
4 comments
Thank the authors for the clarification. However, my main concern is about the sensitivity of the attribution results to this choice. Other reasonable alternatives, such as similarity-weighted averaging, can also estimate the conditional mean. It remains unclear whether such alternatives would substantially change the component-level attributions or affect the main empirical conclusions. I would appreciate further clarification on this point.
Follow-up
(Follow-up: Sensitivity to the summary function) Thank you for the clarification; we now fully understand the concern. We can actually address it both theoretically and empirically.
Theoretical. Let with , that is, a weighted softmax based on similarity, rather than a simple average. Let be the spread of similarities within the retrieved neighborhood. Then, we can bound the weighted-mean and simple-mean difference as This is true because softmax weights satisfy , so . Since both sets of weights sum to 1, it follows that . If (or bounded in general, with a small proof adaptation), we have .
Asymptotically, as the database grows, the retrieved neighbors become arbitrarily similar to the query, so and the weights converge toward uniform. Since each causal effect is a linear functional of the conditional expectation, under strict overlap, for any effect CE, we can show that: which tends to for a fixed as . Therefore, the attributions are equivalent, asymptotically. Intuitively, we are leveraging the fact that K-NN and weighted-K-NN estimators are asymptotically the same.
Empirical. We also tested this empirically. On MIMIC-ED, we replaced the mean by the similarity-weighted average on the same neighborhoods. The shape parameter from above was chosen so that the effective neighborhood size is about or (out of ). Specifically, we report the change (in %), which measures how much the attribution moves:
| Retriever | |||||
|---|---|---|---|---|---|
| Dense | ≈7 | +0.53 ± 0.40 | +0.76 ± 0.47 | +0.25 ± 0.33 | −0.02 ± 0.09 |
| Dense | ≈4 | +0.82 ± 0.73 | +0.69 ± 0.80 | −0.10 ± 0.44 | −0.03 ± 0.16 |
| Lexical | ≈7 | +0.27 ± 0.40 | −0.02 ± 0.45 | −0.30 ± 0.23 | +0.01 ± 0.09 |
| Lexical | ≈4 | +0.44 ± 0.77 | +0.18 ± 0.82 | −0.25 ± 0.35 | +0.00 ± 0.17 |
All changes stay below one percentage point, and most are within one standard deviation of zero; they are small relative to the effects reported in Sec. 3.1, and the main conclusions remain unaffected.
We would be happy to include the result and the experiment in the paper, if the reviewer finds it helpful.
Please let us know if any concern remains! Thank you.
Thank you for the further clarification. I have updated my Soundness score to 3.
Review Response
We thank the reviewer for their comments. We reply to the key points below.
(W1, Q1: Choice of the intermediate predictor) Thanks for the question, it is a good one. Since we see two possible interpretations of the question, we address each separately.
(a) Different retrievers, same output. One interpretation is: two retrievers A and B return different neighborhoods, NKA(q)=NKB(q), yet produce the same output YRAG, while the attributions differ.
First, this setting is unlikely to occur in practice: the LLM conditions on the retrieved neighborhood and its labels, so different neighborhoods mean different evidence, and the output should generally change. In our experiments, for instance, dense and lexical retrieval yield different ΔLLMIE for the same aggregator (Sec. 3).
Second, suppose the outputs do coincide. This would mean that between the two settings, the same output is produced by the LLM while using different evidence; in our view, this does warrant a different attribution to the LLM component. For instance, consider two cases:
The final outputs stay the same, but the mechanisms differ: with a calibrated aggregator LLM (reporting the average retrieved label), setting A would exhibit the retrieval bias δ, whereas B would still have Δret=0. A good attribution should distinguish the two cases.
(b) Same neighborhood, different summary. The second interpretation concerns the intermediate predictor itself.
Yret summarizes the retrieved labels by their mean K1∑i∈NKYi. One may ask whether a different summary of the same neighborhood (e.g., majority voting) would change the attribution, with YRAG unchanged.
The answer is affirmative: a different summary function would change the attribution. However, since the bias signature depends on a score only through its conditional mean (see Sec. 2.1), the role of Yret is to estimate μY(x,z,w)=E[Y∣x,z,w] from the retrieved cases. The mean is the canonical estimator of μY (specifically, it is K-NN estimator): if the retrieved cases match the query on (x,z,w), retrieval is calibrated (Def. 1) and Δret=0 (Prop. 1). A different function, such as majority vote for instance, would estimate 1μY>1/2, and not preserve the conditional mean. This is why the mean function is a principled choice for the intermediate predictor.
Please let us know if the above answers the main concern; thanks again.