司法裁判人工智能化是指機器能代替法官獨立完成某些司法決策。目前主要有兩種進路:一是基于顯式編碼、封閉規則之算法的專家系統;二是基于機器學習算法的預測分析論。法律專家系統雖有多年積累,但限度已顯。大數據算法雖方興未艾,但也同樣難以成功:第一,司法裁判從本質上就無法根據既往數據被預測;第二,機器裁判顛覆了許多司法的基本預設,消解了人的自主權,違背了人類發明人工智能的初衷,而諸如推動類案類判、限制自由裁量等辯護理由都是難以維系的;第三,司法大數據現實上難以支撐算法裁判在技術上的實現。總之,人工智能不能也不應當成為法官那樣的決策主體,更為務實的方向是去發掘其作為輔助工具的價值。

Artificial intelligentialization in judicial adjudication means machines can replace human judges to make certain legal decisions independently. There are two approaches to intelligentializing legal decision making, the first approach is legal expert systems which run on explicitly coded, closed-rule algorithms, and the second one is predictive analytics which develop through machine learning algorithms. Expert systems have developed over the years and relevant limitations have appeared; while big data algorithms are making robust progress, but they are unlikely to be successful as well. Firstly, the adjudication results can not be predicted based on the regularity of big data in nature; secondly, machine adjudication destructs the traditional presumptions of administration of justice, dispels the human autonomy and disobeys the original intention of inventing artificial intelligence, and it is weaker to justify algorithm judgment for its promotion of similar cases being treated similarly or restriction of judicial discretion; lastly, as judicatory system has its particularity, it is difficult to satisfy the needs of learning algorithm technically. Therefore, artificial intelligence cannot and shall not be an adjudicator like a human judge, and what is more appropriate is to tap its potential as an assisting tool in judicatory.