人工知能(AI)の証拠能力に関する一考察(1) : 専門証拠の許容性の観点から

廣島法學 Volume 48 Issue 2 Page 52-31 published_at 2024-10-31
アクセス数 : 112
ダウンロード数 : 31

今月のアクセス数 : 4
今月のダウンロード数 : 1
File
HLJ_48-2_52.pdf 1.02 MB 種類 : fulltext
Title ( jpn )
人工知能(AI)の証拠能力に関する一考察(1) : 専門証拠の許容性の観点から
Title ( eng )
The Admissibility of Artificial Intelligence as Expert Evidence (1)
Creator
YOKOYAMA Yuto
Source Title
廣島法學
The Hiroshima Law Journal
Volume 48
Issue 2
Start Page 52
End Page 31
Number of Pages 22
Journal Identifire
[ISSN] 03865010
[NCID] AN0021395X
Abstract
In recent years, there has been a lively debate about artificial intelligence (AI) in the field of criminal law. This paper examines the admissibility of AI as evidence (AI evidence) in criminal trials, with a particular focus on standards for expert evidence. With the increasing use of AI in legal systems, concerns about the transparency and reliability of AI-generated evidence have become central. The paper highlights that one of the main challenges is the 'black box' nature of machine learning, where the decision-making process of AI is often opaque, making it difficult for judges and juries to assess its reliability. This paper argues for strict expert admissibility standards for AI evidence, as it requires careful examination of the underlying algorithms and data. However, the black box problem complicates the applicability of the standards to AI evidence. Therefore, this paper explores 'Explainable AI' (XAI) and attempts to integrate it into the standards to address this issue.
Language
jpn
Resource Type departmental bulletin paper
Publisher
広島大学法学会
THE SOCIETY OF LAW OF HIROSHIMA UNIVERSITY
Date of Issued 2024-10-31
Rights
許可なく複製・転載することを禁じる
Publish Type Version of Record
Access Rights open access