AI, Trustworthiness, and the Digital Dirty Harry Problem

Abstract

Artificial Intelligence (AI) has been a game changer on many fronts. For the police it offers new ways of carrying out investigation, surveillance, crime prevention and order maintenance. Questions have been raised about the trustworthiness of some innovative AI-driven applications. Under which circumstances and to what extent should the police be permitted to use emergent technology, i.e. use 'dirty' means in order to reach good ends? In this article, this problem is illustrated by a discussion of two emergent technologies, and possible criteria and test regimes for establishing trustworthiness are suggested towards the end of the article.