College of Policing Data Ethics and Data-Driven Technologies APP Consultation: Response from Marion Oswald, Muffy Calder, Kyriakos N. Kotsoglou, Michael Maher, Carole McCartney, Kyle Montague, Claire Paterson-Young and Elizabeth Tiarks on behalf of ‘PROBabLE Futures: Probabilistic AI Systems in Law Enforcement Futures’ Responsible AI UK Keystone Project and Rosamund Powell, Research Associate on behalf of the Centre for Emerging Technology and Security, The Alan Turing Institute

Executive Summary
• We recommend that significant amendments are made to both documents before issuing as APP.
• Use of data-driven technologies and AI must uphold fairness, justice and police duties. While efficiency and productivity are important, a limited focus on these ends (and on metrics such as speed) could mean that errors and issues of validity are overlooked, thus contributing to significant problems such as inadmissibility of evidence at trial.
• The APPs should include more advice about evaluation, how should this be done, who by, and how ‘better’ should be defined. Also, the APPs could give more advice on the incorporation of independent experts in evaluation, including via a data ethics committee, and how to engage effectively with the local community.