Can Foundation Models Talk Causality?
Moritz Willig co-first authorship Computer Science Department TU Darmstadt Darmstadt, Germany Matej Zečević co-first authorship Computer Science Department TU Darmstadt Darmstadt, Germany Devendra Singh Dhami Computer Science Department TU Darmstadt Darmstadt, Germany Hessian Center for AI (hessian.AI) Germany Kristian Kersting Computer Science Department TU Darmstadt Darmstadt, Germany Centre for Cognitive Science TU Darmstadt Darmstadt, Germany Hessian Center for AI (hessian.AI) Germany
Abstract
Foundation models are subject to an ongoing heated debate, leaving open the question of progress towards AGI and dividing the community into two camps: the ones who see the arguably impressive results as evidence to the scaling hypothesis, and the others who are worried about the lack of interpretability and reasoning capabilities. By investigating to which extent causal representations might be captured by these large scale language models, we make a humble efforts towards resolving the ongoing philosophical conflicts.
中文速览
大型语言模型（如GPT-3）究竟有没有"懂"因果关系，还是只会鹦鹉学舌？研究者围绕这一问题展开了系统性实验：他们提出了"因果之上的相关性"这一核心假设——训练数据中本身就包含大量描述因果关系的自然语言表述，使得模型在海量文本中或许能"记住"因果知识，即便它并不真正"理解"因果推断。基于这一假设，研究者用多种措辞向AlephAlpha Luminous、GPT-3和Meta OPT等三个主流大语言模型系统地提问，测试它们能否准确识别已知因果图中的边方向和变量关系，并在常识推理、变量名替换等场景下验证其鲁棒性。结果显示，这些模型在一定程度上确实能给出符合因果图的答案，但表现因模型和问题措辞而有显著差异，尚不稳定。这项工作的意义在于：它为"大模型是否已朝通用人工智能迈进"这场争论提供了一个可量化的切入点，同时也揭示了将语言模型作为因果结构发现工具的潜力与局限，为未来将因果推理与基础模型融合的研究指明了方向。
原文 arXiv:2206.10591;中英对照 + 大白话阅读 https://aha.fim.ai/paper/2206.10591v2