SOURCE-LINKED INTELLIGENCE

An Empirical Study of Counterfactual Self-Explanations in LLMs

arXiv · AI, language, vision and robotics · article · Sep 15, 2026 · UTC

Large language models can easily generate explanations for their own outputs, but such self-explanations are not necessarily faithful to the model's behavior. We study this issue through counterfactual self-explanations, where a model minimally edits an input so that its own prediction changes. Across sentiment analysis and natural language inference, we evaluate ten instruction-tuned models from the LLaMA-3 and Qwen-2.5 families, measuring faithfulness, minimality, and alignment with human-annotated rationales. Our results show that model scale is the strongest determinant of explanation qual

Read original source ↗ Open in workspace

recordType

paper

region

Global