aiqbench analysis
The paper finds that most frontier language models rarely disclose when supplied context actually shaped their answers, and that reasoning-level acknowledgment often fails to reach the final output. Its CREDIT method uses counterfactual labels to supervise final-answer attribution without human annotation, with gains on four open models that partly transfer to unseen information types. For readers, the key caveat is that the results come from specific open models and benchmarks, so the gains need replication before they are treated as a general fix for faithfulness.
Did the Model Use That? Counterfactual Training Boosts Self-Attribution in Language Models
arXiv:2610.35295v1 Announce Type: cross Abstract: A fundamental challenge for explainable and accountable language models is revealing how external information influences their answers. This paper examines whether models are self-attributive: that is, whether they indicate when -- and only when -- their responses to questions are influenced by information provided alongside the question. Furthermore, we explore how self-attribution can be improved through counterfactual training, which explicitly teaches models to distinguish between answers that do and do not depend on supplied information. We found that most of the eleven frontier models rarely attribute influential information in their answers, and the rest attribute non-influential information. Training acknowledgment into reasoning does not fix this: the acknowledgment does not reach the final answer. We introduce CREDIT, which directly supervises final-answer self-attribution with counterfactual labels, without the need for human annotation. CREDIT substantially improves attribution in four open models, including two distilled reasoning models, and outperforms reasoning-level fine-tuning and prompting baselines. Among answers that agree with a user's suggestion, CREDIT distinguishes those changed by the suggestion from those the model would have given anyway. Gains transfer partly to unseen information types and prompts containing sources that support conflicting answers.
Title:Did the Model Use That? Counterfactual Training Boosts Self-Attribution in Language Models
Abstract:A fundamental challenge for explainable and accountable language models is revealing how external information influences their answers. This paper examines whether models are self-attributive: that is, whether they indicate when -- and only when -- their responses to questions are influenced by information provided alongside the question. Furthermore, we explore how self-attribution can be improved through counterfactual training, which explicitly teaches models to distinguish between answers that do and do not depend on supplied information. We found that most of the eleven frontier models rarely attribute influential information in their answers, and the rest attribute non-influential information. Training acknowledgment into reasoning does not fix this: the acknowledgment does not reach the final answer. We introduce CREDIT, which directly supervises final-answer self-attribution with counterfactual labels, without the need for human annotation. CREDIT substantially improves attribution in four open models, including two distilled reasoning models, and outperforms reasoning-level fine-tuning and prompting baselines. Among answers that agree with a user's suggestion, CREDIT distinguishes those changed by the suggestion from those the model would have given anyway. Gains transfer partly to unseen information types and prompts containing sources that support conflicting answers.
29 pages, 5 figures, 20 tables. Code: this https URL
Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
arXiv:2610.35295 [cs.CL]
(or arXiv:2610.35295v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2610.35295 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 308 words