Building alternative AI futures

August 9th marked the nine-year anniversary of when police officer Darren Wilson shot and killed Michael Brown, an unarmed black teenager, in Ferguson, Missouri. Two years after this event, the St. Louis County Police Department adopted the use of HunchLab, a predictive policing system. According to a piece by the Marshall Project, the officers believed the data would “help officers police better and more objectively […] By identifying and aggressively (emphasis mine) patrolling ‘hot spots’ as determined by the software,” the article states, “the police wanted to deter crime before it ever happened.” This continued belief in ‘predictive algorithms’ represents a future that chases the past under the guise of ‘data-driven’ tools.

It feels as though there is a contingent of influential decision-makers who are bound to the idea that ‘predictive algorithms’ actually predict the future. But they don’t: predictive algorithms are rooted in historical data, and can only offer a future that is still tethered to that data. There’s no way we can construct alternative futures out of algorithms trained on the past. Purveyors of predictive algorithms appear to believe more in the past, than in their ability to shape an alternative future.

Nowhere is this dynamic clearer than in policing in the United States. And, in examining predictive policing, we can also uncover the seeds of radically different futures. Let’s dig in, and incorporate lessons from the Untangled Primer along the way, shall we?

‘Predictive policing algorithms’ are spread across over 150 police departments in the US, and have been around since the 1990’s. These systems promise to forecast ‘criminal activity’, and determine where officers should go, and whom they should police. But the training data used for these systems doesn’t actually provide useful insights for those things. First of all, these tools are often trained on arrest data, which means that racially biased police actions directly inform the algorithmic system. Moreover, it’s unclear how some people are considered high-risk at all. For example, the Chicago Police Department developed something called the ‘Strategic Subject List’ to algorithmically predict the likelihood that an individual is at risk of becoming a victim or an offender in a shooting or homicide. But an evaluation of the tool found that more than one-third of individuals on the list have never been arrested or a victim of a crime, and almost 70% of that cohort received a high-risk score. In other words, in an attempt to ‘predict risk,’ the tool actually manufactured it by encouraging an encounter with the police.

So police departments are never exclusively responding to potential crimes, they’re contributing to the production of crime too. This points to the idea that data and technologies often say more about organizations and companies — their organizational structures, interests, and accomplishments — than they do about us or the technology (the fourth Primer theme). The data aren’t ‘predictive’ at all — they’re descriptive and diagnostic of the practices and behaviors of police departments.

Therefore, step one of imagining an alternative future is reconsidering what data can tell us. If they aren’t predictive but diagnostic — then ‘predictive policing’ algorithms offer a nice snapshot of over-policed communities and the practices of police departments.

The second step in imagining alternative futures requires accounting for what these so-called predictive tools leave out. See, one thing that’s weird about these tools is that they don’t include data for predominantly white-collar crimes.

Historically, while social scientists conflated blackness with criminality, white criminality was explained away by structural inequities and poverty. As Khalil Gibran Muhammad documents in The Condemnation of Blackness, in the Progressive era, researchers described whites and immigrants as “a great army of unfortunates” driven “to madness, crime, or suicide” by an inequitable economic system and class oppression. As a result, researchers and reformers advocated for economic solutions (e.g. higher paying jobs, brand new interstate highways leading to white suburbs, etc.) to the problem of white crime. No need to predict it if it’s not considered to be a ‘real’ crime’

One can see this legacy today. Certain predominantly white issues like opioid abuse are seen as the consequence of economic dislocation and treated like a public health issue. Which, it is — but it’s also an example of how some illegal acts are explained away, and not conflated with race. Moreover, some predominantly white crimes aren’t turned into data in the first place, despite the fact that they impact a greater prevalence of people. For example, available studies estimate that approximately 49% of businesses and 25% of households have been victims of white-collar crimes, compared to a 1.06 prevalence rate for violent crimes and a 7.37 prevalence rate for property crime. While the police don’t even bother to predict white-collar crime, researchers and activists have — check out this fun li’l ‘White Collar Crime Risk Zones’ project created by The New Inquiry.

This gets at the ninth Untangled Primer theme, which is the idea that problems are made, not found. Tax evasion data and wage theft data aren’t often included in ‘crime statistics’ because we don’t consider these white-collar crimes part of ‘the problem’ of crime itself. Indeed, the data we select and exclude, the frames we adopt, the boundaries we put on a given system, and the perspective we bring — all these things contribute to what we consider a ‘crime.’ As I wrote in the primer:

How these problems are settled — the process from contestation to commonplace understanding and social acceptance— is about power. Power over who gets to decide what’s at stake; over whose solutions are deemed legitimate.

Step one and step two of building alternative futures give us a clearer depiction of the system we’re trying to change but they don’t account for the underlying dynamics that produce unreliable police data. That’s step three! What, is the underlying culture in police departments which these data emerge from? There are two key areas:

• Power and violence with impunity: Reports interrogating the Minneapolis and Ferguson police departments found that aggressive policing was actually rewarded. This isn’t a new problem; it has thick roots. When the Christopher Commission investigated the LAPD in response to the 1991 Rodney King Beating, it revealed “an informal message that the department conveyed was that confrontational, aggressive policing would be rewarded, even if it resulted in repeated incidents of violence that gave rise to citizen complaints and lawsuits.” The Commission further found that the repeat offenders who generated the most allegations were continually promoted.

• Financial incentives: The Department of Justice issued a report following the killing of Michael Brown which highlighted the “suffering caused by the police treating people as ‘potential offenders and sources of revenue’ rather than as citizens to protect.” The report revealed emails between the Ferguson municipal finance director John Shaw and Chief of Police Thomas Jackson collaborating to boost revenue through fees and fines. In March 2010, Shaw wrote to Jackson, “Unless ticket writing ramps up significantly before the end of the year, it will be hard to significantly raise collections next year. What are your thoughts? Given that we are looking at a substantial sales tax shortfall [caused by the economic recession that began in 2008], it’s not an insignificant issue.” Law enforcement responded accordingly: from 2011 to 2012, revenue generated from municipal fees and fines increased more than 33%, from 1.41M to 2.11M.

The above observations about police culture help us untangle the problem down to its roots. So now, we can identify solutions that address these specific dynamics. Indeed, in her book Viral Justice, Ruha Benjamin highlights the work of Officer A. Cab, a growing abolitionist movement led by survivors of incarceration, which advocates for a number of policy changes that would directly address the system dynamics identified above, such as: ending qualified immunity, terminating civil asset forfeiture, breaking the power of police unions, requiring malpractice insurance, and changing the funding incentives that shape police practice.

With all this in place, the final step is developing alternatives. As Audre Lorde famously put it, “For the master's tools will never dismantle the master's house. They may allow us temporarily to beat him at his own game, but they will never enable us to bring about genuine change.” In other words, we can’t use the tools of policing to change policing, we must develop alternative approaches to community safety. As scholar Dorothy Roberts put it in an interview with Ruha Benjamin in Captivating Technology:

“The answer, to me, is the leadership and vision of strategizing and thinking and organizing of women of color, who want to abolish those structures going into the algorithms. What would it mean to give agency to the survivors of state and private violence to imagine and implement the best way to deal with it.”

There are a number of experiments trying to do just that, as outlined by Ruha Benjamin:

• There are Community HEAL Circles that are led by survivors of violence, which center restorative justice, and allow those formerly incarcerated to “explore topics such as trauma, shame, accountability, structural and generational violence, and impacts on our bodies.”

• There’s the group Sister to Sister, led by mostly immigrant women who were often harassed by police when they called to report domestic violence, who in turn started “workshops, street theater, vigilance committees, trainings, so that men, women, kids in the community, understood how to deal with domestic violence.”

• There’s the Creative Interventions initiative which “hosts conversations in cities around the country about how everyday people can avoid 911 calls and choose community-based alternatives.”

Those who look at a list like this and dismiss it as a set of naïve ineffective interventions are exhibiting the exact kind of reaction that upholds the status quo. As Arundhati Roy wrote, “The first step towards reimagining a world gone terribly wrong, is to stop making war on those who have a different imagination.” This framing actually provides a helpful new first step in our program to build alternative futures, so the complete set of steps might look something like this:

1. Stop rejecting alternative approaches as unrealistic — they’re only unrealistic to the extent that you accept the status quo.

2. Reconsider what the data is actually saying by viewing algorithmic systems as diagnostic and descriptive, rather than predictive.

3. Account for what’s not counted in the data.

4. Account for the underlying dynamics that make the data, and then align solutions to those dynamics.

5. Experiment with alternative systems that are not dependent on the previous system, making new data, untethered to the past.

Reliance on current systems has us inadvertently perpetuating a dangerous fiction: that we are ‘predicting the future’. Really, we’re just making ourselves dizzy chasing the past — like a dog chasing its own tail. We can’t keep looking at the future like it’s some dark unknown; the only way to break out of this is by letting go of the false dream of ‘prediction’ and start actually taking control of the future by shaping it ourselves.