Can Artificial Intelligence Make Police Officers Better?
Law enforcement is under constant scrutiny for the force that is used in the execution of their jobs, and shouldn't they be? When you carry a variety of weapons and have the authority to use them, you also carry the responsibility. The key for every law enforcement agency is to help officers develop and maintain the skills to de-escalate situations without the use of force when possible. In the police academy and once a year after that, the Los Angeles Police Department requires all sworn officers to be “tested” in the application of deadly force using the Force Option Simulator (FOS). Officers stand in front of a large screen and are presented with a pre-recorded video where officers have responded to a call for service. Prior to the start of the scenario, officers are given a replica firearm tethered to a machine. As the scenario unfolds, officers must talk to the screen as they would in a real-life situation and make the right choices regarding the use of force.
FOS sounds like a good idea, the only problem is, it doesn't work...well. Here is how it operates today. As you talk to a screen, the screen does not respond back to your directions. It doesn’t give much training value when the conversation is one way. Having been in an officer-involved shooting, I can tell you there are many things you have to consider when de-escalating a situation—tone of voice, content of your message, and building rapport. Unfortunately, the current training has no mechanism to train or test this. Fortunately the LAPD like many police departments across the country, do provide in-service training where an officer pretends to be a suspect while the trainee officer talks to him or her. This type of training is vastly better then FOS. Unfortunately this type of training is extremely expensive do to the amount of time and resources required to provide it and the scenarios that are included in the training are limited.
What if artificial intelligence could be used to create a better FOS program? The key is creating a dynamic interaction between the trainee and a virtual person that can be responsive to the 3 important factors of de-escalating situations: tone, content and rapport. The trainee could be presented with a scenario complete with a computer generated “suspect” able to react to what the trainee says and does, much like Amazon's Alexa responds to your music requests. The new increases in processing power and storage are allowing AI powered technologies like computer visioning to create solutions like the one described above. Not only could the standard human FOS program be improved but it could include an endless variety of scenarios limited only by the time allocated to trainees to train. Valuable tools like this can help officers vastly improve their use of de-escalation techniques. Ironic that computers may be the key to helping people become better at dealing with people.
Author: Mike Miller - Director Public Safety and Threat Intelligence at AIXIA Global