同案同判的关键是相似性判断的证成，这需要借助类比论证来实现。运用论证型式解析类案判断，相似性判断可被分解为数量相似性、相关相似性和决定相似性，分别对应“哪里相似”“何种相似”“为何相似”的批判性问题。人工智能技术可通过量化计算的方式对案例的相似度进行精确表达，在一定程度上实现类案的智能推送。而无论是相关相似性还是决定相似性的判断，均需借助法官的价值判断来完成。对此，法律人工智能将价值判断转化为经验判断，能够为法官的类案判断提供参考。但是，其无法代替法官选择类案判断的实质标准。因此，案例系统作为智能工具，在辅助法官进行类案判断时有明确的限度，不能依此消除合理的差异化判决。

人工智能, 类比论证型式, 数量相似性, 相关相似性, 决定相似性

The key to uniform judgments in similar cases lies in the assessment of similarity, which requires the use of analogical argumentation. By employing analytical schemes for argumentation, similarity assessment can be decomposed into quantitative similarity, relevant similarity, and determinative similarity, corresponding to critical questions of “where similarity lies” “what kind of similarity it is” and “why it is similar” respectively. Artificial intelligence technologies can precisely express the similarity of cases through quantitative calculations, thus achieving intelligent recommendations for similar cases to a certain extent. Whether in the judgment of relevant similarity or determinative similarity, judicial value judgments are necessary to reach a conclusion. In this regard, legal artificial intelligence transforms value judgments into empirical judgments, offering limited guidance for judges’ analogical judgments. However, it cannot replace the substantive standards for selecting analogical judgments by judges. Therefore, while case systems serve as intelligent tools, aiding judges in analogical judgments, they have clear limitations and cannot eliminate reasonable differences in judgments.

rtificial intelligence, analogical argumentation schemes, quantitative similarity, relevant similarity,  determinative similarity