RoboCop: AI’s Impact on the Future of Police

On November 30^th, the San Diego police department received approval from supervisors to deploy robots with lethal capabilities in emergency situations, making it the second California city to publicly discuss the use of remote lethal force after Oakland in October 2022. The decision generated immediate backlash which pushed supervisors to put the program on hold following a vote on December 6^th.

San Diego and Oakland may have halted the use of “killer robots” for now, but they have been used previously in the United States after Dallas police used a robot to kill a mass shooter in 2016. The mass shooter had already killed five police officers and injured dozens more. The robot was equipped with an explosive, then moved to a wall behind the shooter, then the bombs exploded killing the mass shooter and slightly damaging the robots arm, marking the first usage of a “killer robot” by U.S. law enforcement.

Robots in domestic law enforcement are not unusual in U.S. police departments, and many have operated robots for close to a decade but they have been used to inspect suspicious packages or deliver items during hostage negotiations without risking police officers and until 2016 not for lethal purposes.

Although “killer robots” grab our attention because they force us to imagine a dystopian future where law enforcement is similar to “Robocop”, the future of artificial intelligence in law enforcement is likely to be much more mundane. However, it is due to this that makes current AI in law enforcement more dangerous than ‘killer robots’, one only needs to look at the two photos below and choose which they feel more threatened by.

Is it the robot equipped with a gun or the robot that slightly resembles a trash can? We will assume that the ‘killer robot’ is more threatening to you, however, despite being armed the ‘killer robot’ can only function with a human operator otherwise it is just a 400kg chunk of metal with a gun. However, the friendly trash can operate completely autonomously, is constantly watching, is able to scan your license plate, your mobile phone, and, identify, then report individuals that it deems suspicious. It’s ability to identify and report individuals is done according to its AI, and while AI does not have inherent racial bias that does not make it a perfect assessor of who and what is a threat. This is vital to understand as current AI is far from perfect and in some cases misidentifies individuals with dark skin far more than those with light skin, highlighting the importance of the data that AI is trained with and the underrepresentation of people of colour who make up the majority of the world.

Although racial discrimination is a particular problem of U.S. law enforcement, the use of robots is not isolated to the U.S., China has been using autonomous robots in its airports and roads, and used robotic dogs and drones to police Shanghai during lockdown.

Full Lockdown in Shanghai, this is how they broadcast announcements.

Robot Dog + Speakers#Shanghai #COVID #Lockdown pic.twitter.com/5kJdLrnL8p

— Jay in Shanghai 电动 Jay (@JayinShanghai) March 29, 2022

Dubai launched the UAE’s first robot officer which has built in cameras that stream to police command centres and was deployed in malls and tourist attractions and during Expo 2020 drones were used to secure the event and had the capability to detect 92 types of incident such as fires and traffic incidents.

Compared to ‘killer robots’ these are less threatening to us because they look disarming and they are marketed as such, with companies boasting of the amount of social media engagement that their robots produce. But their current ability to constantly watch and monitor paired with continued innovation in the field of AI raise serious concerns regarding the future of privacy, surveillance, and the relationship between police and policed.

Returning to San Diego, the backlash against ‘killer robots’ was focused on the further militarization of the police. U.S. police are able to equip themselves with military grade technology through the Department of Defence’s 1033 Program which sees surplus military equipment transferred to local police departments, military equipment such as the Explosive Ordnance Disposal Robot that was used in Dallas in 2016.

Therefore, to better understand the future of robots and AI in local law enforcement, one only needs to look at their current military applications to have insight into the possible technologies that police will adopt, as warfare is usually the source of innovation in these fields.

Israel was able to assassinate Iran’s chief nuclear scientist with the assistance of a AI equipped gun capable of firing 600 rounds a minute and operated thousands of kilometres away via satellite. This is not new technology for Israel as it already uses it to police its border it is however the first known use of it in a foreign operation. This AI-assisted gun joins the combat drone in the long list of robots capable of killing however, where a drone can be spotted and shot down in the sky, a simple machine gun is much more difficult to find.

Previous use cases of ‘killer robots’ show that a human pulled the trigger and not the robot itself meaning that the machine was incapable of killing on its own, this changed in 2021 when a report from the UN highlighted the use of Turkish autonomous drones that were programmed to attack targets without requiring permission from a human operator marking the shift from human operated lethal robots to autonomous killer robots.

This is the technology currently being deployed in warfare and it is a matter of time before it is adopted by local law enforcement to some degree. Currently 61% of people oppose lethal autonomous weapons systems according to a poll, although it is a majority of people there is a large amount of variance across countries and the number will likely erode as the world faces new threats.

This raises a series of questions for the future of AI in law enforcement. How will human police officers be utilised in the face of the automation of some of their most traditional functions, how should we educate and train our law enforcement in the use of these machines, and how will the relationship between human and robot police be maintained?

Police robots may seem cute now and provide a fun photo opportunity, and they are marketed by the companies that produce them as such but when we are aware that they are constantly monitoring will we be as eager to take a selfie with them, and how cute will we find them when they have the ability to kill and view us simply as data, and are we willing to continue putting human lives in danger when a robot can take their place?