Justice and AI
by Elena Connarty, Communications and Events Officer
No industry or individual wants to miss out on the potential that AI can offer to improve and accelerate processes. For the public sector, often constrained by limited funding and resources, AI could become an essential tool. In recent years, the police have been exploring how this technology could enhance their efforts. The use of AI in policing has, however, not been without scrutiny.
In this blog we explore the use of AI in criminal justice and hear from experts Jo Callaghan and Alessandra Fassio. Accompanying this blog is our podcast with Jo Callaghan, where we discuss this topic and her integration of AI in her recent crime novel series. Read on to learn more, and listen to our podcast for Jo’s perspectives.
Recent Developments in AI Policing
The most well-known and discussed development of AI use in policing is facial recognition technology. In 2023, Policing Minister Chris Philip highlighted the importance of using innovative technologies such as AI to aid crime prevention and monitoring.1 Using live facial recognition, this technology captures live crowd images and compares them to a database of known subjects, potentially identifying suspects and alerting nearby officers.1 This could be seen as a tool for identifying dangerous individuals and creating a safer environment for all.
However, these systems have faced backlash. Studies have shown that facial recognition systems are less accurate in identifying individuals with darker skin tones, leading to a disproportionate impact on minority communities.2 This technology can therefore lead to racial profiling, and wrongful arrests due to inaccuracies and biases in the AI algorithms. Identifying and controlling these biases in AI/data systems, allowing for adjustments and improvements to be made, will be essential if such technology is to be used in the future.
Predictive policing is another tool being trialled across the UK, using algorithms and data to forecast where and when crimes such as burglaries and violence are likely to occur.3 It has also been employed to predict the behaviour of current criminals, estimating the likelihood of reoffending. This could inform decisions about the next steps in a criminal's rehabilitation program.3 The tool is expected to ease resource pressures and enhance public safety.
Regulating AI in Criminal Justice
To ensure public safety and trust in policing, the public must have confidence in their police force. Integrating AI into policing without transparency about when, where, and how it is used can lead to public distrust.
Transparency is key to building this trust. If technologies such as facial recognition are employed, there must be clear notices informing the public, similar to those used with CCTV. Data should not be stored on these systems unless it matches someone on a watch list, and if no match is found, the data should be automatically deleted. Furthermore, the judicial process must remain involved, ensuring that individuals have the right to fair representation in court.
AI systems often rely on data that reflects existing biases and inequalities, which can lead to discriminatory policing. These risks can reduce trust in the justice process, particularly among those who are frequently subject to discrimination. To ensure AI tools function effectively in policing, strong regulations must be established to govern their use and the data they process. Additionally, the outputs of these systems should always be reviewed by a human to monitor and correct any biases.
Exploring AI in Policing Through Fiction
To explore the topic of Justice and AI further, our Communications Manager Gordon Johnstone sat down with Jo Callaghan, a Senior Strategist specialising in AI research in the workforce. Jo Callaghan, now also an Author, is applying her research in AI to that of policing.
Having studied AI’s role in the justice system in depth for her book series she was able to provide us with some fascinating insights.
AI and Intuition
During their conversation, Jo emphasised that AI should be seen as an augmenting tool rather than a replacement for core roles within jobs. Specifically, in the context of policing, she noted that while instinct and judgment are crucial in detective work, these qualities can be subjective. AI can introduce more evidence-based decisions, enhancing the investigative process. However, Jo pointed out that justice also requires empathy, an attribute AI lacks. Her novel, In the Blink of an Eye, explores the benefits and drawbacks of AI integration in policing, encouraging readers to consider these aspects.
Ethical AI Policing
Jo also addressed the negative press surrounding previous AI policing efforts, such as facial recognition. She highlighted the flaws in training data biases, particularly against minorities, and stressed the importance of fair data training and input to mitigate these biases. Jo advocated for public input in AI usage decisions, arguing that AI is currently used predominantly for commercial benefits rather than the public good. To truly benefit the public, it is crucial to understand what people need and trust.
AI Characters in Literature – Locke
Jo Callaghan has been instrumental in creating an AI character that feels realistic and grounded in possible future technology. In her novel, the character Locke uses deep learning to analyse data much faster than humans. However, the book makes it clear that humans are essential throughout the process, particularly in making final judgments beyond the data. AI lacks the human emotions necessary for working in justice and policing, where understanding each unique case requires looking beyond mere data.
This fascinating discussion with Jo Callaghan is available to listen wherever you get your podcasts, just search for The Scottish AI Alliance, or access here.
The Future of AI in the Justice System – Insights from Alessandra Fassio
Senior Data Ethicist at the Ministry of Justice UK, Alessandra Fassio, shared her perspectives with Gordon Johnstone on the future of AI in the justice system over the next 5-10 years. Alessandra is optimistic about AI's potential to enhance efficiency, effectiveness, and empathy in the justice system, particularly for vulnerable service users. She emphasised the importance of maintaining a people-centric approach in the justice system, even as we embrace technological advancements.
Alessandra believes that AI can significantly reduce administrative and desk tasks, enabling frontline staff to dedicate more time to interacting with service users and creating a more people-focused service. By automating routine tasks, AI can free up valuable time for staff to provide personalised support and care, enhancing the overall quality of service in the justice system.
AI can also play a vital role within prisons by enhancing inmates' digital literacy and skills. This can facilitate smoother reintegration into society and help reduce reoffending rates. By providing inmates with the necessary skills and knowledge, AI can contribute to more successful rehabilitation and reintegration efforts.
Alessandra highlighted that AI will not only influence policing methods but also the nature of crimes. The rise of issues such as revenge porn and deepfakes will require a more proactive approach, requiring legal adjustments to keep pace with technological advancements. To address these challenges, the government must ensure that new technologies are proportionate to criminal activity and avoid more invasive policing methods than necessary.
Overall, Alessandra Fassio envisions a future where AI enhances the justice system's efficiency, and effectiveness, that will allow for more time to focus on a people-centric approach.
Conclusion
The future of AI in policing and criminal justice is a topic of significant interest and debate. AI has the potential to enhance the justice system but only if implemented thoughtfully and ethically.
The insights shared by experts Jo Callaghan and Alessandra Fassio highlight the complexities and possibilities of AI in this field. From addressing biases in data to ensuring human oversight in decision-making, there are many factors to consider as we move forward.
To explore these topics further and gain a deeper understanding of the future of AI in policing, listen to our podcast featuring discussions with leading experts. Search for The Scottish AI Alliance wherever you get your podcasts, or follow this link.
References
1. https://www.gov.uk/government/news/police-urged-to-double-ai-enabled-facial-recognition
2. https://www.aclu-mn.org/en/news/biased-technology-automated-discrimination-facial-recognition
3. https://post.parliament.uk/ai-in-policing-and-security/