Harnessing the Power of Artificial Intelligence in Law Enforcement: A Comprehensive Review of Opportunities and Ethical Challenges

Introduction

In the 1940s, the first computer in the modern sense was invented, and it wasn’t long until the young academic Alan Turing began to think about the future potential of the computing machine (Anyoha, 2017). Despite the very limited performance of the computer at the time, he wondered if computers in the future would not have to think in a way that lived up to or surpassed humans. Humans use information and reason to solve problems and make decisions, so why shouldn’t machines be able to do the same? Alan died in 1956, but two years later, Marvin Minsky picked up the baton and co-hosted with John McCarthy the Dartmouth Summer Research Project on Artificial Intelligence. They invited well-respected researchers from various fields for an open discussion about artificial intelligence (AI). Henceforth, artificial intelligence became an accepted term. Research continued, and the field developed, but computers’ storage capacity and computing power were limiting factors, and interest in AI cooled (Anyoha, 2017). During the 80s, AI saw another boost due to increasing research funding, and combined with the rapid development of performance in the 90s, it could be concluded that AI was here to stay (Kaynak, 2021).

Today, we live in a world where AI has found its way into more and more areas of society and completely changed how we work. Politics, education, healthcare, business, and law are just some societal areas that have undergone major changes. An area that has become highly relevant recently is law enforcement, where traditional routines still dominate (Kaynak, 2021). The majority of the police work that takes place is only done after the crime has occurred, which means that large resources are required in the form of personnel, money, and time. In addition, the possibilities of solving the crime decrease as time passes. With the help of AI, the police can perform more predictive police work, i.e., to predict where, how, and when crimes will be committed to be better equipped for calling out or preventing crimes (King et al., 2020). The hope is that with AI’s help, one can analyze material from video cameras in real time. Another prospect is being able to read social media to identify patterns and behaviors that may be indications that crime is about to happen. Successful implementation of this could save many lives Freeman (2020), save enormous resources, and minimize the risk of human error that can occur in today’s criminal investigation (Zufferey et al., 2022). These are just some of the areas where it is suggested that AI can bring about major improvements.

Herrmann (2023) believes in its AI report that the use of AI in the judicial system must take place in such a way that it considers the general principles of human rights, democracy, justice, and the prevailing legislation. To succeed with this consideration, authorities within the legal system must work to achieve the requirements of fairness, accountability, transparency, and explain ability. In recent years, these requirements have been developed based on consensus within the AI community about what algorithms are considered to possess to instill trust. Fairness means that algorithmic decisions should not show discriminatory or unfair tendencies. The requirement calls for all AI systems to be scrutinized to ensure they comply with the right to non-discrimination. Accountability means there must be clear regulations for who is responsible for an autonomous system’s decision. Transparency means that there must be clear answers about the goal of using AI in a certain context and what parts the AI consists of. It could, for example, be about which data is used. The last requirement, explain ability, is closely related to transparency but is more focused on the person affected by a certain decision being able to understand the algorithmic decision in non-technical terms (Herrmann, 2023).