Abstract: ChatGPT is a Natural Language Processing (NPL) Model driven by a new generation of artificial intelligence technology. The article discusses in depth the Causal Rules, Causal Representations and Causal Explanations of ChatGPT around causality. By analyzing the underlying logic in ChatGPT’s causal cognition and learning, the following three understandings are obtained: First, ChatGPT uses correlation as its underlying causal rule, driven by Deep Neural Networks(DNNs), Large Language Model（LLM）and statistical analysis methods, and realizes the embedding of causality in artificial intelligence in the form of probability. Second, to realize the simulation of human’s ability of causal cognition, causal language and causal inference, ChatGPT has a representation similar to human’s intelligence in the generation of natural language. Third, relying on the two paths of Computational Causal Explanation and Causal Emergent Explanation generated by intelligence, ChatGPT realizes the input from symbols and information to the output of "Data Knowledge", and acquires new capabilities in an emergent way, realizing artificial intelligence The "Cause and Effect Revolution" in the field of artificial intelligence constructs a blueprint for the future development of artificial intelligence.

摘要： ChatGPT是新一代人工智能技术驱动的自然语言处理模型。文章围绕因果性深入讨论了ChatGPT的因果规则、因果表征与因果解释，通过分析ChatGPT因果认知与学习中的底层逻辑得出以下三个认识：一是ChatGPT将相关性作为其底层因果规则，在深度神经网络、大语言模型与统计分析的驱动下，以概率形式实现了因果性在人工智能中的嵌入。二是通过实现对人类因果认知、因果语言与因果推理能力的模拟，ChatGPT在自然语言的生成中拥有了与人类智能相似的表征。三是依托智能生成的计算因果解释和因果涌现解释两条路径，ChatGPT实现了从符号与信息的输入到“数据知识”的输出，并以一种涌现的方式获得新的能力，实现了人工智能领域的“因果革命”，构建出未来人工智能发展的蓝图。