LLM自我解释是否有助于用户预测模型行为？通过语用扰动评估反事实可模拟性
Do LLM Self-Explanations Help Users Predict Model Behavior? Evaluating Counterfactual Simulatability with Pragmatic Perturbations
AI总结：
本文通过语用扰动评估LLM自我解释对用户预测模型行为的帮助，发现自我解释能提升模拟准确性，但效果受扰动策略和评判者能力影响。
摘要：
大型语言模型（LLMs）可以生成口头化的自我解释，但先前研究表明，此类解释可能无法可靠地反映模型的真实决策过程。我们探讨这些解释是否仍能帮助用户预测模型行为，将其定义为反事实可模拟性。使用StrategyQA，我们评估人类和LLM评判者在有无访问模型的推理链或事后解释的情况下，预测模型对反事实后续问题答案的能力。我们比较了LLM生成的反事实与基于语用的扰动作为构建测试用例的替代方法，以评估解释的潜在有用性。我们的结果表明，自我解释在LLM评判者和人类中均能一致提高模拟准确性，但收益的程度和稳定性高度依赖于扰动策略和评判者能力。我们还对人类用户在预测模型行为时所写的自由文本解释进行了定性分析，这提供了证据表明，访问解释有助于人类在扰动问题上形成更准确的预测。
英文摘要：
Large Language Models (LLMs) can produce verbalized self-explanations, yet prior studies suggest that such rationales may not reliably reflect the model's true decision process. We ask whether these explanations nevertheless help users predict model behavior, operationalized as counterfactual simulatability. Using StrategyQA, we evaluate how well humans and LLM judges can predict a model's answers to counterfactual follow-up questions, with and without access to the model's chain-of-thought or post-hoc explanations. We compare LLM-generated counterfactuals with pragmatics-based perturbations as alternative ways to construct test cases for assessing the potential usefulness of explanations. Our results show that self-explanations consistently improve simulation accuracy for both LLM judges and humans, but the degree and stability of gains depend strongly on the perturbation strategy and judge strength. We also conduct a qualitative analysis of free-text justifications written by human users when predicting the model's behavior, which provides evidence that access to explanations helps humans form more accurate predictions on the perturbed questions.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
