摘要： 生成式人工智能对法律人的挑战引发了学界的广泛讨论，这里的挑战指的是对法律职业需求的侵占。法律职业需求中，对疑难案件的处理能力是法律人的核心竞争力，其中“等置模式”是主要处理模式，若生成式人工智能可模拟这种模式，那么就会对法律人构成真正挑战。生成式人工智能的主要技术原理，是自注意力机制和预训练大语言模型。预训练大语言模型与法律人的知识体系具有同构性，并可通过自注意力机制进行概率推理，这使其模拟等置模式成为可能，进而构成挑战。为因应这种挑战，需要发挥法律人的主体性，策略有三：重视法学方法论的教育，转向以法律证立为重心，从概率推理走向法理分析。

关键词: 生成式人工智能, 等置模式, 预训练大语言模型, 概率推理, 法律人主体性

Abstract: The challenge posed by generative artificial intelligence to jurists has sparked widespread discussion in academia. This challenge refers to the encroachment on the demands of the legal profession. The ability to handle complex cases is the core competitiveness of jurists, with the “equivalence model” being the primary processing mode. If generative artificial intelligence can simulate this mode, it will pose a real challenge to jurists. The main technical principles of generative artificial intelligence are self-attention mechanisms and pre-trained large language models. The pre-trained large language model has an isomorphic structure with the knowledge system of jurists and can perform probabilistic reasoning through self-attention mechanisms, making it possible to simulate the equivalence model and thus pose a challenge. To address this challenge, it is necessary to leverage the subjectivity of jurists, employing three strategies: emphasizing the education of legal methodology, shifting the focus to legal justification, and moving from probabilistic reasoning to jurisprudential analysis.

Key words: generative AI, equivalent pattern, pre-trained large language model, probabilistic reasoning, jurists’ subjectivity