让大模型回溯：如果指令不同，结果会怎样？
Should I Have Expressed a Different Intent? Counterfactual Generation for LLM-Based Autonomous Control
- 用因果模型建模用户-模型-环境闭环，支持反事实推演
- 通过校准生成的反事实结果集，包含真实结果概率超95%
- 适用于想优化指令但怕试错的智能控制场景
基于大语言模型（LLM）的智能体可将用户高层意图转化为环境中的计划与行动。然而，在观察到结果后，用户常会思考：如果当初表达得不一样，结果会不会更好？我们提出一个框架，使这类反事实推理在基于LLM的自主控制场景中成为可能，并提供形式化的可靠性保证。该方法将用户、基于LLM的智能体与环境之间的闭合环交互建模为结构因果模型（SCM），并利用测试时扩展，通过概率反向推断生成多个候选反事实结果。通过离线校准阶段，所提出的置信反事实生成（CCG）能生成包含真实反事实结果概率极高的结果集。我们在无线网络控制案例中验证了其性能，显著优于直接重执行的基线方法。
原文摘要 · Abstract (English)
Large language model (LLM)-powered agents can translate high-level user intents into plans and actions in an environment. Yet after observing an outcome, users may wonder: What if I had phrased my intent differently? We introduce a framework that enables such counterfactual reasoning in agentic LLM-driven control scenarios, while providing formal reliability guarantees. Our approach models the closed-loop interaction between a user, an LLM-based agent, and an environment as a structural causal model (SCM), and leverages test-time scaling to generate multiple candidate counterfactual outcomes via probabilistic abduction. Through an offline calibration phase, the proposed conformal counterfactual generation (CCG) yields sets of counterfactual outcomes that are guaranteed to contain the true counterfactual outcome with high probability. We showcase the performance of CCG on a wireless network control use case, demonstrating significant advantages compared to naive re-execution baselines.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。