YEAR 2020 No 2 Vol. 22

ISSN 2182-9845

Machine Learning Evidence

Fernando Silva Pereira

Keywords

Artificial Intelligence; Machine Learning Evidence; Evidence; Civil Procedure Law.
 

Abstract

Machine learning is a field of artificial intelligence that gives computers the ability to learn without being explicitly programmed, posing the problem of using the outputs of deep learning software as evidence in a judicial process. Focusing on Civil Procedure Law, this article aims to reflect on this problem, from the point of view of the admissibility and weight of such an evidence, giving close attention to the north-American experience, where the problem of the use of scientific and technic evidence has been largely discussed.

Table of contents

1. Introduction
2. Artificial intelligence and “machine learning”
3. The problem of admissibility of evidence
3.1. Introduction
3.2. North American Supreme Court jurisprudence on the issue of the admissibility of technical and scientific evidence and Rule 702 of the Federal Rules of Evidence
3.2.1 The general acceptance test, or Frye test
3.2.2 Further developments and entry into force of the Federal Rules of Evidence
3.2.3 Decision in the Daubert case v. Merrel Dow Pharmaceuticals, Inc. and court gatekeeper
3.2.4 Rule 702 of the Federal Rules of Evidence and the Daubert/Kumho’s criterion
3.3. Rule 702 and the resulting proof of “machine learning software”
4. The problem of assessing evidence
5. Conclusion
Bibliography