Breaking the Chain: A Causal Analysis of LLM Faithfulness to Intermediate Structures 事件

Breaking the Chain: A Causal Analysis of LLM Faithfulness to Intermediate Structures arXiv:2603.16475v2 Announce Type: replace Abstract: In schema-guided reasoning (SGR) pipelines, LLMs produce explicit intermediate structures -- rubrics, checklists, or verification queries -- before committing to a final decision. SGR is increasingly adopted because it promises controllability: practitioners expect to inspect, edit, and override these structures to steer the outcome. But does the promise hold?