Originally Posted Here
The use of artificial intelligence (AI)- based predictive analytics and Big Data analytics within police departments across the globe has been a contentious issue since the technologies were introduced, especially when it comes to just how much autonomy AI-powered systems should have.
In fact, the UK Home Office recently called off an AI project that was intended to predict whether people would commit their first violent offence with a gun or knife over the next two years. The system was found to be wildly inaccurate and was rejected by a group of experts due to ethical issues.
When it comes to predicting crime offences, there is far more that makes up the picture than the historical data alone. Crime is caused by many different and sometimes unrelated factors. There is almost always not a direct correlation between a crime instance and the various situations, thought processes and data sets that are generally available for AI machines to compute on. This means that the actionable insights, predictive analytics and recommendations provided by the AI machines could only be treated as an accessory to police decision-making, rather than the sole driver of decisions.
While police departments must tread carefully on this terrain, there are still ways for them to put predictive analytics to good use ethically. There are several solutions out there that allow police teams to use data and generate actionable insights into decisions around resource deployment and allocation while remaining cognizant of potential biases and ethical problems. Here are the top ways in which the police can use predictive analytics to augment and empower their day-to-day decision-making.
Predictive analytics powers resource deployment
With past and present real-time public and police data on crime hubs, police forces can make better- informed decisions around deploying their resources each day.
When planning assignments of officers, police professionals can understand which areas require more staffing on the ground, the kind of back-up support to have available and what type of vehicle or mode of transport is best. For example, a predictive analytics tool could predict a rise in crime levels in a certain area due to an upcoming event, such as a fair or concert. This event would lead to more traffic, which correlates with higher crime levels. By predicting higher traffic levels, police forces can anticipate in advance that crime is likely to peak and deploy additional resources, with the right tools, at the right time and in the right place.
In addition, data-powered resource deployment decisions mean police officers can justify why they are patrolling certain areas, using certain tactics, and be held accountable for instances of over- policing that may be influenced by human bias.
Predictive analytics allows for better infrastructure planning
Predictive analytics platforms can use data from multiple sources, cross-reference it and augment police department decision-making around infrastructure planning. For example, a new pub or bar might open in an area that does not usually see much late-night activity. Based on past trends, police can anticipate a rise in crime in this area as it will attract more people who are consuming alcohol late at night.
They can plan for this and consistently allocate more resources to support policing infrastructure in the area, such as street cameras, and potentially even build new police stations around these upcoming hubs.
The reverse case works for AI-powered infrastructure planning, too. If a certain area is experiencing an economic boom and there is an influx of residents that can cause crime levels to drop, police departments can divert resources elsewhere. Such predictive analytics models provide valuable insights into upcoming urban migration trends.
Various police departments in the US have been using geographical prediction tools that use data modelling to forecast risk in specific locations across cities. With the tool, it is easier to pinpoint
“Too many police forces invest in data analytics tools without even scratching the surface of their potential power”
at-risk locations and the conditions under which crime increases or decreases and support individual police officers on the beat.
Predictive analytics helps establish criminal patterns
Armed with predictive analytics tools that can mine through and process thousands of datasets at a time, police can now identify patterns and correlations that they might have otherwise missed by human eyes.
For example, police departments in the US have drawn from data on crime locations and travel trends showing how criminals were commuting from one area of a city to the other to commit robberies. The destination area had recently seen the development of a number of upmarket apartment complexes, which were easy and fruitful targets for offenders who were willing to travel to that side of town.
Knowing this, police can better track and monitor criminal activity in those areas.
Uncovering patterns like these from data might have been the job of a full army of police professionals before. Now, with predictive analytics platforms, police forces can get these insights immediately and make informed decisions on how to act on them.
How to effectively leverage predictive analytics tools
Too many police forces invest in data analytics tools without even scratching the surface of their potential power. This is largely down to a lack of training and know-how in making the most of the software.
If a force is going to invest in a comprehensive solution, like that of Palantir or IBM, it should have an in-house team that understands the capabilities of the tool and how best to put it to use. However, there are out-of-the-box solutions such as Safera, which take the onus off police forces to extensively train staff or employ technical teams.
No matter the solution, just like within other organizations that are leveraging data analytics for the first time, police departments must promote a mindset and cultural change across the board for teams to truly appreciate and value what data analytics tools can offer.
In addition, teams must also remain mindful of the bias that analytics tools can generate. There have been multiple instances of unethical use of AI solutions within police forces. It is crucial not to treat data-powered predictions critically, so as not to profile certain segments according to things like race, nationality or socio-economic status. Opting for platforms that do not ‘see’ factors such as race or income status will help avoid bias that often occurs within police profiling.
Given the misuse of AI tools in recent years on the part of law and order officials, it is understandable why some police forces may be hesitant to adopt the technology.
What is important to remember, though, is that analytics tools are as biased as the data which they are fed with, and rarely is an algorithm fed with perfectly fair data. This means keeping humans in the loop at all stages of the analytics process to prevent instances like biased profiling.
By instilling a data culture across the team, and choosing the right solution according to their needs and capabilities, police forces can leverage the best of data analytics and power their decision-
making to new heights.