Bias and AI in Judicial Application

EasyChair Preprint 15598

Abstract

This paper analyzes the international developments for the containment of bias and the causes and consequences of unobserved bias in IT forensics used in investigative and expert activities. AI offers new and interesting possibilities for this, although not all types of bias are relevant when using AI. It should also be noted that several biases can occur one after the other, which further increases the erroneous nature of the investigation results. LLMs as special AI systems can certainly offer various areas of application in expert witness work or investigative work, although it should be noted that LLMs have a strong tendency to hallucinate. Finally, international developments in the use of AI in court will be surveyed and the problems that are to be expected, especially when using LLMs.

Keyphrases: Beweismittel, IT-Forensik, Künstliche Intelligenz, bias

@booklet{EasyChair:15598,
  author    = {Thomas Hrdinka},
  title     = {Bias and AI in Judicial Application},
  howpublished = {EasyChair Preprint 15598},
  year      = {EasyChair, 2024}}