Will AI Be the Public’s New Crime Fighting Superhero?
When we think of crime fighters, superheroes such as Batman and Spiderman probably come to mind. The reality is, sadly, a lot less exciting. Usually, security staff sit in rooms fitted out with banks of monitors to surveil the displayed video feeds from cameras showing our cities, airports and public spaces.
But humans have trouble concentrating for long periods of time and can be easily distracted. This is where AI is already performing a greater role in the campaign against crime and terror. Machine learning techniques can already recognize a wide range of objects in real time. With tens of cameras feeding into a security office, an AI can sort the images and highlight the feeds that require immediate review by the human team.
Of course, not all crimes are captured on camera. Even video evidence of a crime’s occurrence may not provide the required information to charge all those involved with orchestrating the crime. This is where offline analysis of video can help. Using deep learning techniques, recordings can be analyzed turning the scenes captured into a standardized text description. This text is termed metadata. Whereas video requires gigabytes of storage for a few minutes of recordings, a textual description may require just a few tens of words.
For example, a subject under surveillance may be captured by different surveillance systems in different locations across several days. The subject may have been carrying a red suitcase on these occasions. By comparing the AI generated metadata from these disparate surveillance systems, occurrences of red objects, specifically suitcases, can be quickly determined and the associated video frames can be identified along with the time and date.
As security analysis broadens its reach, it is quite possible that other data sources will be added to the mix. Other AI skills, such as facial and car number plate recognition, can also be turned into metadata. When combined with other information, such as calls to the police by the general public, a bigger picture can be quickly built out of seemingly unrelated events.
In 2017 the government of Singapore announced it was investing around 22% of its tech budget into security, a sum of around $528m. Much of this will go towards the development of AI and analytics at their first Government Security Operation Centre that focuses on cyber threats. The outcomes from this type of threat detection will likely compliment other evidence collected for counterterrorism.
AI will probably not be our sole crime fighting superhero as there are too many facets and angles to be considered that require human cognitive skills. However, there are many opportunities for AI to support us in the fight against crime and terror attacks.