Current And Future Use Of AI In Law Enforcement

Artificial Intelligence isn’t just a buzzword anymore. It has become part of our lives, and its uses and applications are growing every other day. Even the public sector, which usually is a late adopter of new technologies, has come onboard this new train.

Law enforcement, in particular, has seen the advantages different AI technologies can offer to their work and has started to integrate them into their workflow and daily routines. But there is much more to come.

Task automation: AI as a workhorse

Due to the increasing importance of media files in police investigations, current police cases have an increasing amount of digital files to be analyzed. Videos from mobile phones, computer files, sound recordings, voice messages from chat applications…the list is almost endless. But, in contrast to other types of digital files, like documents or PDFs, it is not possible to search directly a certain information in a video. It has to be watched by someone. And that takes an inordinate amount of time for a human being, as there can be hundreds of videos in a case.

This is one of the areas where artificial intelligence shines. Modern analytics systems are able to find almost any kind of information in media files due to the improvement of artificial vision, object recognition and face biometrics. It is simply a matter of feeding the hundreds or thousands of media files to a AI analyzer, which will work through them and find specific sounds, words, faces, cars, etc.

Case of law enforcement

On top of that, an analyzer does not tire after long work hours and does not make errors

What is more, unlike human officers, these systems can work 24/7 which speeds up investigations considerably, as more evidence is found in less time. On top of that, an analyzer does not tire after long work hours and does not make errors.

This is why police work can be helped greatly by AI Analytics, as it frees police officers to do high value work, instead of endlessly watching videos on a computer of listening to audio recordings, in search of evidence. As soon as the system finds what the officer has specified (a face, name, number plate, object, etc.) it sends an alarm to the officers’ phone, so he or she can take a look and decide what to do with that information.

Despite AI being touted as a danger to many workplaces, in the case of law enforcement, it is a valuable tool to help police do more, with more accuracy, and in less time, freeing officers from the repetitive and boring work of checking mountains of evidence in search of clues.

The AI Analyzer landscape

The field of AI analytics is expanding constantly and new types of analysis are being discovered that may be helpful to police or intelligence. The most used analyzers today are probably Automatic License Plate Recognition (ALPR), object and face recognition. But OCR is equally useful not just for reading documents, but also signs and logos that may appear in pictures or videos, to help identify a location.

And for audio (as in interception or surveillance recordings) there is speech-to-text (S2T), translation, Speaker ID, audio fingerprinting (AFP) and natural language processing (NLP) which is able to extract sentiment from what is said. All of these have their application in daily police work and can save time in investigations or make them possible in the first place. But looming on the horizon are new possibilities, which we will discuss later and that are even more powerful. So the field of AI Analytics is, by no means, a closed one.

Generative AI in law enforcement

LLMs will be an important part in all those tasks related to investigate large sets of documents

For the past year, the term “Generative AI” has become part of our general vocabulary, although most of the time we just say ChatGTP, Copilot o similar. The large language models (LLM) use Deep Learning and different AI strategies to, amongst other things, analyze and summarize vast amounts of information, in order to generate a short report with the highlights. This can be of use for Law Enforcement in all those cases that have great amounts of documents that may contain evidence. Again, this is a case like the above, where AI helps speed up operations by doing the grunt work much quicker than any person could. 

The difference is in the Deep Learning part. The model can be tuned to specific needs (like financial crimes, for example) and will get better over time when dealing with specific sets of documents. Thus, LLMs will be an important part in all those tasks related to investigate large sets of documents during a case.

What the future holds

As with all technology, it is virtually impossible to predict what the future will hold. Because any breakthrough can upend complete sectors, as demonstrated by ChatGPT not that long ago. However, there are several promising AI technologies in the pipeline, some of which are already being tested and perfect around the world:

• Behaviour analysis: as facial recognition systems get better and better, they are not only able to recognize faces, but also facial expressions. This means that AI systems could be assisting during interrogations, to evaluate the truthfulness of what is being said. Combined with the analysis of small voice inflections, they can be a non-invasive “lie detector”.
• Robotics: already in use by many police forces around the world, robots are going to be ever so important. Particularly the autonomous kind, which is able to do missions on its own, without a human behind the controls. This, combined with swarm technology, could be an incredible help in disaster areas, where time is of the essence in locating victims.
• Predictive policing: thanks to pattern analysis, predictive policing, which has been to the test several times already, will be an important part of police work, to figure out where to send units or concentrate surveillance efforts.

In short, AI has much more to offer, and we are going to see and incredible evolution of this technology applied to law enforcement, over the next years.