Predictive Policing and Artificial Intelligence – book review

More relevant today than in 2021 

When I was asked to review ‘Predictive Policing and Artificial Intelligence’, published in 2021, before the release of OpenAI’s ChatGPT in 2022, I mused to myself whether it would be similar to conducting a review of the printing press while sat behind my MacBook Pro (other computers are available). But I could not have been more wrong. 

Prominent criminologists Ken Pease and John McDaniel have brought together a diverse set of experts. They have delivered a wonderful mix of theoretical and practical tips for policing to consider as we wrangle with the best ways to deliver AI technologies.  

A balanced, thoughtful exploration of AI in policing 

Broken into separate chapters considering different technical or ethical aspects of AI in policing, the book delves into how AI and predictive algorithms are reshaping policing, drawing on experiences from the UK, US and Australia. It offers a nuanced analysis of AI’s potential to enhance policing while addressing ethical and operational challenges. The editors balance optimism about AI’s capabilities with caution about its responsible implementation. 

Bias and big data – harnessing AI’s potential 

Part 1, ‘Bias and big data’, explores how AI can revolutionise policing by improving analysis and predictive capabilities. It introduces concepts like machine learning, neural networks and risk assessment models within criminology. A key chapter, written by Pease and McDaniel, discusses choice architecture – how AI tooling will increasingly influence police decision-making. The authors emphasise that AI must be implemented thoughtfully to avoid reinforcing historical biases in crime data. Part 1 largely presents AI as a tool that can enhance policing effectiveness if designed responsibly. 

Ensuring accountability and public trust 

Part 2, ‘Police accountability and human rights’, focuses on the importance of governance and considering ethical concerns. The book poses crucial questions about transparency in AI decision-making and accountability for AI-driven outcomes.  

There is a standout chapter, written by Dr Alexander Babuta, Director of the Centre for Emerging Technology and Security, and Honorary Lecturer at University College London, and Professor Marion Oswald, Professor of Law at Northumbria University and Senior Research Associate at the Alan Turing Institute. It deconstructs machine learning and its applications in policing, making recommendations for governance and oversight. Part 2 is an excellent reference point for anyone involved in the deployment of AI tooling, particularly those who support predictive policing or decision-making.  

An essential read 

Despite its academic focus, the book is largely digestible, offering valuable insights for police leaders, data scientists, policy makers and practitioners. Each chapter stands alone yet forms a cohesive narrative that is both thought-provoking and practical.  

While it definitely is not a light bedtime read, it is an impressive collation of some of the world’s most prominent thinkers on data-driven technologies in policing. If I was to read it again, I would focus on the introduction and pick key chapters of interest, taking my time with each chapter. Alternatively, it could be used as a reference book to support, for example, those charged with setting up a data or AI ethics forum, where chapters from part 2 would be incredibly helpful.  

Fundamentally, ‘Predictive Policing and Artificial Intelligence’ points the reader towards how transformative AI will be for policing. This is a transformation that requires intentional, responsive and thoughtful oversight and governance. This in turn will deliver both better outcomes for the public and enhanced confidence in our service. 

• This article was peer reviewed by Russell Holloway, Solutions Architect, British Transport Police