UK police department trials AI system to expedite cold case investigations

UK Police Department Trials AI System to Expedite Cold Case Investigations

The Avon and Somerset Police Department, responsible for policing parts of South West England, is currently trialling an AI-powered system that promises to revolutionise the review of cold cases. Developed in Australia, the platform, known as Soze, is designed to perform extensive investigative work in a fraction of the time traditionally required.

In initial tests, Soze has demonstrated its ability to scan and analyse vast quantities of data from various sources including emails, social media accounts, video content, financial statements, and other documents. Remarkably, the AI system processed evidence from 27 complex cases in approximately 30 hours. This workload would have required around 81 years of human labour, underscoring Soze’s potential as a significant force multiplier, especially valuable for law enforcement agencies facing personnel and budget constraints.

Gavin Stephens, chairman of the UK’s National Police Chiefs’ Council, expressed enthusiasm for the potential applications of an AI system like Soze. “You might have a cold case review that just looks impossible because of the amount of material there and feed it into a system like this which can just ingest it, then give you an assessment of it,” he told Sky News. “I can see that being really, really helpful.”

However, the implementation of Soze in everyday policing encounters the critical issue of accuracy. The accuracy rate of the Soze platform has not yet been disclosed, which raises concerns. AI models have a known propensity to produce incorrect results or hallucinate information, a phenomenon where the AI generates plausible-sounding but entirely false outputs. This aspect necessitates rigorous validation to ensure reliability and to avoid potential miscarriages of justice.

In addition to Soze, another AI initiative mentioned by Stephens involves the creation of a database cataloguing knives and swords used in violent crimes. The aim is to assist in identifying suspects involved in such attacks. The rollout of these AI tools is seen as imminent, but their effectiveness and the possibility of inherent biases need careful consideration.

Historically, AI applications in law enforcement have been fraught with challenges. Previous AI systems designed to predict the likelihood of reoffending have been criticised for inaccuracies and racial biases, notably against Black individuals. There have been instances where AI facial recognition led to wrongful arrests. These concerns were notably highlighted by the US Commission on Civil Rights, which recently criticised the use of AI in policing due to the significant risks of bias and error.

Despite the promising efficiencies presented by AI, the reliance on machine-generated analysis remains controversial. The perception that AI systems are inherently infallible because they are machine-based is misguided. The data these systems are built on and trained with originate from human input, which can often be biased or erroneous, thus inheriting similar issues.

The Avon and Somerset Police Department’s trial of the Soze AI platform marks a significant step towards integrating advanced technology in police work. However, ensuring the accuracy and fairness of these AI systems remains paramount. As these technologies begin to play a larger role in investigations, rigorous scrutiny and ongoing assessment will be essential to their success and to maintaining public trust.

Source: Noah Wire Services