返回全部动态
基于LLM的因果推断鲁棒性压力测试
原标题：Stress-Testing the Robustness of LLM-Based Causal Inference
AI 摘要
该论文提出一个覆盖模型规模、因果图复杂度、提示策略、元数据质量、数据质量和不确定性敏感度六个维度的系统性压力测试框架，用于评估基于大语言模型的因果推断。研究发现LLM存在明显的乐观偏差，倾向于高估因果关系导致精确率普遍偏低；传统评估指标具有欺骗性，纳入不确定性后性能明显下降；模型规模扩大带来的收益脆弱，元数据质量下降时性能急剧退化。
以上摘要由 AI 生成，可能存在误差。事实请以原文为准。
正文节选
Stress-Testing the Robustness of LLM-Based Causal Inference Abstract Why do some of the world’s most powerful Artificial Intelligence (AI) systems still struggle to reason about cause and effect? Causal discovery, the ability to discern the underlying drivers of a system, is fundamental to scientific inquiry and has traditionally relied on structured data and statistical methods. Recently, Large Language Models (LLMs) have been explored as an alternative paradigm for causal inference, leveraging
发布时间：2026-11-05 13:00
抓取时间：2026-11-05 13:12
来源机构：arXiv