Making the most of AI and new technology for policing
October 17, 2019
Across the UK, police leaders are working with AI and new technologies to keep their communities safe and secure. Yet, from South Wales to San Francisco, a quick glance through the global reporting about police use of AI and new technology, shows the challenges that come hand in hand with potential benefits when it comes to new technologies such as facial recognition. So, how can police use AI and new technology wisely and responsibly?
Changing societal expectations, coupled with speed of data and delivering proactive policing are the key issues identified in our Policing in a Networked World report, and AI and new technology are at the heart of all three. Having seen predictive policing pilots, the roll out of body worn video, facial recognition trials, through to the implementation of an ethics panel and development of single online home for policing, there are five key issues I would highlight to any police body stepping into this space:
Focus on the ‘why’
Policing needs to be clear about its strategic intent in using AI or new technology. Is success testing technology and thinking? Or is it a reduction in crime in an area? Having clarity on the end goal for using a certain technology is essential from the start.
Consider how governance and oversight will work and who is accountable, including potentially establishing an ethics panel. The UK is ahead on its use of police ethics panels and they are proving to be good for UK policing. Police leaders know if they don’t get the ethics of this right it damages legitimacy and trust. Our Responsible AI toolkit provides a practical framework for decision makers.
Be clear and open in internal and external communication about what the technology is, and as importantly isn’t. If it is AI and the algorithm learns and develops, then explain it. If it doesn’t and what you have done is automated an existing process be clear on that.
The quality of data counts
UK experience shows new systems can often be hampered by the poor quality of legacy data, for example large quantities of duplication or an inability to link people across systems. Use the rollout of new technology to drive data quality. This should include involving practitioners in the choice of datasets to support any algorithm, as they should understand any challenges with the datasets you are using. If you are going to use commercially available datasets make sure you have a similar level of understanding about the base data as you do with police systems.
Learn from others
Finally, policing is at its best when it looks outward, and this is one of those areas where it helps to look, listen and explore far and wide. Two sectors where real learning on the ethical challenges of technology are available are the health sector, where AI is now used in processes like cancer screening, and financial services, where AI and process automation is transforming services.
While the challenges are real, the potential AI and new technology offers to transform modern policing means we must find ways to ensure policing can harness new technology in a fair, responsible way that helps keep communities safe and secure.