Wired Magazine reported that San Francisco has banned private law enforcement’s use of facial recognition technologies. It should be noted that this doesn’t affect private sector use of the same technologies. You can still unlock your phone with a wink and a glance and Google Glass junkies can safely look up who you are. It also doesn’t affect federal law enforcement. The event comes at the decision that any benefits that may come from this technology do not offset the risk of unreasonable, weaponized surveillance. In other words, legislators watched the movie Minority Report one time and felt appropriately uncomfortable.
To consider some of the mindsets involved in this ban, it can be important to remember the sizeable effect that fast-paced, advanced startup technologies have in the Bay Area. That part of the country in particular deals with the bleeding edge of artificial intelligence and data processing capabilities as a matter of fact. This gives them a vantage that offers strong faith in the capabilities of the tools and tech and a sane wariness of the speed at which it evolves. If you’ve ever been automatically tagged in a Facebook book photo because the platform recognized your face, it’s like that except there’s a cop browsing the photos.
Let’s Talk Benefits
There are certainly benefits to technologies like facial recognition in law enforcement. However, does the benefit outweigh the risk or potential cost? Is this an illusion of safety or can lives really be saved here? Nationwide there are about 5 or 6 daily unsolved missing person cases in the US. To be able to easily track and locate individuals that have disappeared, or quickly find suspects of crime seems easily worth it. Personally, if one of my daughters went missing I’d ask the cops to turn on all cameras right then.
Let’s further note that there are the human resources freed from conventional forms of investigative leg work through the use of these technologies. This means that suspects of violent crime could be more quickly identified and brought to justice. However, while facial recognition is powerful it isn’t flawless. The likelihood of mistakes is probably lower than that of suspect lineups, but certainly not zero. Further, dataset bias may become responsible for a digital form of racial or class profiling.
That being said, it is probably more beneficial than harmful to utilize these technologies limiting to the two use cases above. Human-machine hybrid partnerships has already proven exceedingly effective in fields like medicine in which IBM Watson powered machines are hugely accurate at identifying cancerous cells, outperforming solo human counterparts at avoiding false negatives. Could such a futuristic buddy cop combination prove better at finding missing persons or identifying and apprehending suspects of violent crime or theft?
AI Police State
It is necessary to ask if there is a real threat from local law enforcement having access to these technologies. A police state is generally regarded as an unpalatable thing in the United States. It is defined as “a totalitarian state controlled by a political police force that secretly supervises the citizens’ activities” [Google Dictionary]. Whether or not facial recognition in the hands of a local police force leads to totalitarianism is debatable (it’s unlikely), but what *is* intuitive is that modern machines’ abilities to connect the dots can be used to create fictitious cases against the citizenry with ease.
Algorithms could be designed that make it easy to detain innocent citizens, and due to phenomena like AI bias, they wouldn’t even need to be designed intentionally for this purpose. For example, predictive analytics can be used to determine the likelihood that an individual, let’s say you, might commit a crime. It isn’t important whether or not you did or would, what is important is that as an individual for whom a court has reasonable cause to suspect you could be searched and seized. Further, this is a scenario without personal malice. Using facial recognition, cases and arrests could be made against any sort of rabble-rouser, from activists to political opposition.
Private vs. Public Implications
One might ask what the difference is between public institutions using readily available civilian technology and private ones, especially when those private companies already track over 2 billion of the world’s population. Well, it might be hard for you to do (hard being an understatement), but you *can* opt out of using commercial platforms like Facebook or Google. You can also make a purchasing decision on the devices you use and their capabilities. Really, one of the key things free market capitalism provides is the ability for consumers to change the sway of corporate decisions by making something unfashionable. You can also politic for government intervention when corporate powers become too mighty.
Opting out of municipal government is difficult, and it is unethical to force someone to do so. Further, where creating a new privacy-friendly Facebook competitor might be close to impossible, I assure you creating a new government is much harder. Unfashionable products end with new products. Unfashionable governments end in street riots. While we still believe in avoiding unreasonable search and seizure, we need to keep sacred what the term “unreasonable” means. Finally, outside of policymaking, we have very little say as to where surveillance cams and processing servers can and can’t go. You’d essentially be asking the system to change itself on your behalf. This is a reasonable request for a citizen to make of their local government and law enforcement, but not in a world where rabble-rousers can have an AI-generated rap sheet produced based on their arbitrarily tracked behaviors.
Effort and Resistance
One could argue the very good point that the methods being used are no different than the investigative methods already implemented now. The only difference is the speed and coverage of investigative data processing. However, it is exactly this speed that truly concerns most people. At least at the subconscious level. Part of the comfort we feel with current policing is that of its finite available effort and attention. In other words, only so much policing can be done in a given amount of time by a given amount of people. All laws aren’t infinitely enforceable.
The effectiveness at which a large deployment of facial recognition database connected public cameras could potentially search and identify individuals breaks down that barrier of effort. There are people who remain innocent simply because of the effort and the consequences of committing a crime. Similarly, there are crimes that go unsolved or even unaddressed because of the effort and resources that would be required to chase after the criminals. That resistance of effort is a type of safety net. When government powers become strong enough to overcome such resistance of effort, the citizens, out of distrust, tend to themselves become the resistance.
Imagine the not entirely unlikely scenario of having enough data to connect an individual to multiple minor infractions. Jaywalking, downloading a copy of the latest Marvel movie, or simply placing them in proximity to enough illegal activity to raise an eyebrow. This collection of information could be used against that individual. It is in fact how Al Capone was eventually put away. While he couldn’t be arrested for racketeering, they found sufficient infraction in tax evasion to do the job. Now I’m not saying that the average person is a modern-day Al Capone, or that there is enough interest to put people in prison for 127 counts of jaywalking, however, a threat does exist so long as there is any financial incentive in fining or incarcerating citizens. If you have ever been caught by a traffic light cam, just imagine this phenomenon extended pretty much anywhere a camera could sit.
Changing the Narrative
The above scenario is a bit sensationalist and it is honestly unlikely for it to go completely out of control. However, so long as mostly innocent people feel unprotected and even at odds with municipal law enforcement, a sense of opposition will remain. Honestly, despite the creepy 1984 feelings that even I get from the thought of this technology, I’m in favor of such forms of surveillance so long as the intellectual and constitutional rigor are put forth to define scenarios in which abuse is likely to occur and systems to provide oversight and enforcement of their use and implementation.
However, before I am truly comfortable with this, the narrative between citizens and law enforcement needs to be rewritten. The financial incentive to imprison, and privatization of prisons will need to be eliminated, and the relationship between cops and civilians, criminal and innocent alike, will need to be healed with trust and the necessary checks and balances. Without these actions, the prologue of our present conversations will foreshadow the worst for the world that is careless with this future.
Digital Surveillance is Inevitable
And it is the future. In the end, putting this technology in place is inevitable. The methods would not only become more effective and less expensive, but it will also naturally find its way to be integrated into every tool and device used in both the law enforcement and civilian sectors. That’s why I feel we can’t forestall the conversation that should be had right now, especially not before the systems become so complex and powerful that we won’t know exactly where to begin the dialogue.
This type of enforcement does not need to be all or nothing. Rather it should be designed around wise concepts with compassionate intent and structured to fight against its own abuse. As I mentioned before while this is problematic, it is ultimately a matter of saving lives. Or at least it should be. The danger comes when it doesn’t become about that. When it *does* become about power, or about manipulation. The best way to avoid this is to begin by putting all cards on the table and earnestly having the discussion. This way, corrupt and manipulative use of publicly tracked facial records are treated with earnest and deliberate care and it remains for the good of the citizenry.
What I intentionally did not mention throughout this post is the question of whether or not this is Constitutional. If you are creating a corpus of information that effectively tracks the activity of citizens, that makes it easy and extremely effective to track the history of an average citizen, isn’t this, in fact, unreasonable surveillance? Couldn’t you argue that every citizen is being actively and even aggressively tailed by digital police? Any benign series of records could be brought up to weave a suspicious narrative. It’s as good as having a warrant with your name on it just waiting for a signature. This discussion needs to be had against the intent of our Bill of Rights. At no point was the Constitution drafted with the expectation of omniscient and omnipresent sentinels.
Finally, a lot of this conversation is actually moot. The expense and sophistication of the implementations I have been describing and fascinating about simply don’t exist. There simply aren’t that many cameras, that much bandwidth, or that much data. It is simply precautionary for the sake of creating a complete thought experiment. Much like magical forensics used by TV cops, the type that can enhance blurry images and accurately complete a DNA test within a couple of hours, most of this is fantasy. However, eventually, it won’t be. The rate of technological advancements is exponential, not linear. All the more reason to have a safe and sane conversation in the present.
What do you think? Should digital facial recognition be banned from municipal law enforcement? Or is it a safe and useful tool in the hands of investigators and police? Please let’s begin the conversation.
Last week or so, they solved a homicide by seeing the victim on a red light camera. Some people were upset that they were able to just pull those to see people driving by. I figure, I’m in public so what am I hiding? I rather a killer be off the streets.
Yes, while I see both sides of the story, I have to lean on the side of progress. To some extent, we have to use the tools we have to make the world better and safer. It’s essentially a human imperative. Every tool can be weaponized, but that doesn’t mean we ban hammers, we simply put checks and balances in place to control access and curb abuse.