In focus: Avon and Somerset's experiment with crime-prediction

WIRED has published the results of a comprehensive investigation into a sprawling predictive analytics programme operated by Bristol City Council and Avon and Somerset Police. Launched in 2016, the Think Family Database stored sensitive data, including police intelligence, mental health records, and school meal details, on nearly half a million individuals. Officials used machine-learning models to assign risk scores to adults and children to map threat levels.

The report highlighted that at least 23 predictive models were developed, including an Offender Management App that holds data on 300,000 people to identify crime risks. However, the programme was found to lack transparency. At least two other risk-scoring models were quietly abandoned after council staff deemed them untrustworthy. Independent analysis of over 36,000 performance scores also indicated poor predictive performance. 

Despite inspectors and independent reviewers warning that these systems could undermine public trust, the UK article concludes that the direction of travel is clear. At the start of the month, the government announced that it is proceeding with a national rollout, backing a new PoliceAI body to deploy tools across 43 forces.

Training Announcement: The BCS Foundation Certificate in AI examines the challenges and risks associated with AI projects, such as those related to privacy, transparency and potential biases in algorithms that could lead to unintended consequences. Explore the role of data, effective risk management strategies, compliance requirements, and ongoing governance of the AI lifecycle and become a certified AI Governance professional. Find out more.