AI在刑事證據法之研究：以美國聯邦證據規則為核心

A Study on the Use of Evidence Generated by Artificial Intelligence in Criminal Proceeding: U.S. Federal Rules of Evidence

現代社會中，人工智慧(AI)的判斷或預測結果逐漸成為人類決策的重要參考。雖然在刑事訴訟的論罪階段，尚無將AI之判斷結果，使用來證明被告有罪或無罪的案例；但隨著技術廣泛應用，訴訟一造主張以AI的判斷為論罪證據，恐是司法審判須面對的新型態資料。本文聚焦於具備機器學習過程並能獨立提出判斷結果的AI，以美國聯邦證據規則為基礎，試以現行法律框架探討可能的適用問題。本文認為，此類AI證據屬於數位證據之一種，但無論視為物證或人證，均有適用法則之扞格。因此，需調整證據法的適用模式，以因應此新型態之證據。

In the wake of rapid developments in artificial intelligence (AI), judgments or predictions made by AI are being used in place of human judgment or as a reference source. Although there have been, as yet, no cases in which judgments or predictive data generated by AI have been used as the basis for criminal charges in the fact-finding stage, it is anticipated that, given current trends and the already extensive application of AI, there may come a time when the judiciary must confront evidence of this new type in criminal proceedings. This article focuses on AI equipped with machine learning processes and capable of independently rendering judgmental results as the primary subject of discussion. It aims to examine applicability issues within the existing legal framework, using the United States Federal Rules of Evidence as the subject for the application of evidentiary rules. This article contends that evidence derived from AI falls under the category of digital evidence. However, the current framework governing evidence law will encounter challenges treating artifacts of AI as real or testimonial evidence, necessitating adjustments addressing this form of evidence.