Concerning Data
Is Artificial Intelligence “the biggest event in human history”? If you think so then you are in good company. Stuart Russel, founder of the Centre for Human-Compatible AI and Professor of Computer Science at the University of California, Berkeley, has asserted that “at some stage we should expect the machines to take control”. Quite an unnerving prospect for operational Police decision makers whose remit can include everything from resolving challenging crime situations through to officer care and budget management.
In recent times we have witnessed the explosion of data generating devices. Mobile ‘telephones’ have progressed from 1G to 5G, phones and drones carry cameras and big data is churned out by the simplest of IoT devices and the most complex of satellites. If 90% of data has been generated in the last two years(1) and quantum computing is on the horizon, then how can we find meaning from the data we already have and how do we prepare for what is to come?
It is reasonable to expect that Police effectiveness and therefore citizen safety will improve if organisations with responsibility for public safety(2) can work in a more integrated way and share data and information. However, apart from the obvious governance challenges, integrated cross-organisational decision making based on common situational awareness presents risk and challenge, even if all data are complete and accurate. Decision makers are vulnerable to being overwhelmed with data which can be a distraction rather than an aid, leading to challenges of confused or conflicting priorities, and not forgetting data protection and privacy barriers, especially where multiple organisations are involved.
From a Police perspective, how can we find meaning from data that exists within the Service? With the advent of data science skills and techniques, more informed and earlier decisions can be made by experienced policing professionals. There may still be no substitute for experience but there is opportunity to complement intuition with data and information. Hexagon’s Command and Control capability features Machine Learning algorithms which can help anticipate events and allocate Police resources more efficiently. In Coventry, Oracle recently predicted crime(3) volumes and future hotspots at the recent BAPCO conference. This information can be used today to help allocation of future officer and vehicle resources based on incident probability. Further, the information can be assessed in order to determine the impact that future decisions will have on crime outcomes.
Stop and Search data analysis(4) can also help consider efficient resource allocation. In Durham, 27% of stop and searches resulted in arrests, 14% of which were resolved in the community. Was there an action in the community that could be predicted to avoid these incidents? In Gwent no action was taken following 96% of searches - predictive data is already able to show the impact on operational policing priorities that the same police resources could have had elsewhere. Thanks to the vast quantities of data already collected and processed by Police organisations, resource allocation decisions can be taken today to prepare for crime predictions 6 – 12 months in the future.
Although it is tempting to concentrate on the immediate benefits such as allocation of Police resources to emergency incidents, the longer-term consequences of AI adoption need to be considered: Should AI be allowed to choose between dispatching resources to either one emergency incident or another or will human ethical judgements always be needed5? When AI and quantum computing are the norm can decisions about future Police resource profiles be trusted to machines based on future crime predictions? We need to take time together to consider such matters.
(1) National Policing Digital Strategy. (2) Eg Police Services, Fire and Rescue Services, Ambulance Services, Social and Health Services, Counter Terror organisations and Armed Forces. (3) Eg: violent and sexual offences, possession of weapons anti-social behaviour, and burglary. (4) www.data.police statistics from December 2014 to January 2020. 5 And if so is it possible for Stuart Russel’s statement be qualified: “at some stage we should expect the machines to take control”?
Author:
Jan Thompson, Oracle
Georgie Morgan
Georgie joined techUK as the Justice and Emergency Services (JES) Programme Manager in March 2020, then becoming Head of Programme in January 2022.