Lentis, Predictive Policing

Background

The first symposium on predictive policing was in Los Angeles in 2009. 1] Predictive policing seeks to track trends in crime and try to better prevent it. Though predictive policing has come to focus on technological solutions, a major precursor to more modern Artificial Intelligence (AI) is the collection of crime statistics into a database to track crime. According to Charlie Beck, police chief of the LAPD, “The predictive vision moves law enforcement from focusing on what happened to focusing on what will happen and how to effectively deploy resources in front of crime, thereby changing outcomes.”[1] Common data analysis strategies used in predictive policing include [data mining, crime mapping, geospatial prediction, and social network analysis. These strategies are used to predict high-crime locations and aid in police decisions. Results of predictive policing are increased surveillance and patrolling of areas deemed to be high risk and increased surveillance of individuals marked at risk of being involved in criminal activity. [2]

CompStat

CompStat, a performance management system and the direct precursor to today’s AI predictive policing, was an early software that allowed police to create a database of information, allowing them to make their own predictions of future crime from the accumulated data. By collecting this data, it leads departments to “place a strategic focus on identifying problems and their solutions.” According to Garry McCarthy, Superintendent of the Chicago PD, “CompStat is not a solution. It’s a method to obtain solutions.” [3] This is a difference between CompStat and the AI systems that followed it as these systems took what CompStat did a step further, using the data to guide police resources and directly providing solutions.

PredPol

PredPol was one of the earliest AI software to be used in predictive policing and it was created back in 2010 by UCLA scientists working in conjunction with the LAPD. [4] PredPol works by identifying locations where serious crimes are likely to occur over a period of time, based on the data input of the town or city's past criminal activity. This is then frequently used to direct officers to specified locations for their patrols. PredPol claims to be twice as accurate as human analysts but this has not been confirmed by independent study. [4]

Racial Bias

The American Civil Liberties Union (ACLU) and other civil rights organizations have raised serious concerns about racial biases in the results from this software.[4] These concerns have lead to lawsuits raised against departments employing the use of predictive policing technologies. Independent studies have also found that targets of predictive policing technology tend to align closer to race rather than actual criminalized activity. One such simulation study tracking PredPol targets for policing against actual estimated drug use by race found that drug use by white residents was over triple that of the percent PredPol targeted for policing. [5]

Police Departments

A large number of major police departments employ predictive policing, including the NYPD,[6] and other police departments have only ended their long running programs in 2020, including the Chicago PD[6] and the LAPD.[7] Some departments feel that by using predictive policing programs, they can more accurately police high crime areas, and say that even if the programs do not directly predict crimes, just knowing that there is a predictive policing program in place will be a deterrent for criminals.[8] Despite multiple complaints of racial biases in predictive policing by advocacy groups[8][9] and citizens,[8][10] departments continue to use the programs, with one LAPD spokesman saying "It is math, not magic, and it is not racist."[11]

At this time, no police departments or police affiliated groups have come out against predictive policing.

Companies that Create Predictive Policing Software

Palantir company logo. (Image: Palantir, Public domain)

Companies that sell predictive policing technology, such as Palantir or PredPol, often market their solutions as being “data driven” and reducing the work of police officers.[12][13] Palantir aggregates data about individual arrest records or citations in order to target specific people who they consider likely to commit crimes, while PredPol compiles location data in order the identify hot spots where crimes are likely to occur.[10][13][14]

Transparency issues tend to plague these programs, and multiple police departments have been sued in efforts to find out how they work.[6][10][15] Most of these lawsuits are driven by claims of racial biases within the programs, which police departments deny, although they often refuse to release data to back up their claims.[6][15] Neither PredPol nor Palantir has directly responded to racial bias complaints, although PredPol has a section on their website where they reiterate that their data is not based on demographics.[16]

The Communities Being Policed

Community members report feeling targeted and over-policed, with one person saying that asking "how often do I see police in my area is like asking me how many times do I see a bird in the day.”[17] Because predictive policing is based on previous police data, many communities argue that it simply reinforces existing racial biases in policing, creating a "feedback loop" where communities that have historically been over-policed will continue to be over-policed.[6][8][17] This is supported by studies which show that information in police databases can be heavily racially biased, since police are more likely to stop people of color rather than white people; this leads to predictive policing programs producing heavily biased outcomes.[10]

Advocacy Groups

Advocacy groups include both large organizations like the ACLU and the Brennan Center, and smaller community based ones like the Stop LAPD Spying Coalition (SLSC).[6][9][18] These groups do not support predictive policing, with the ACLU saying that American policing “is systemically biased against communities of color and allows unconscionable abuses of police power,” and that giving them a supposedly neutral algorithm allows them to deny allegations of racism.[9] The ACLU and the Brennan Center have both sued police departments to try and gain information about their predictive policing programs.[6][15]

With LAPD's predictive policing program being shut down, many members of the SLSC feel hopeful about future fights, with cofounder Hamid Khan saying "Predictive policing has roundly been discredited. This [decision] was clearly [the result of] the organizing that was done. This was clearly the community rising up.”[7][19]

The Los Angeles Police Department has a long history with racism and abuse (the beating of Rodney King in 1992), and predictive policing software it injected into its structure in the early 2010’s has helped revitalize a push against the organization.

LASER Program

Beginning as a research project in 2009, and first being put into use by the LAPD 2011, the Los Angeles Strategic Extraction and Restoration program is a point automation system used to flag citizens in Los Angeles who may be likely to commit a crime. [20]A target is assigned and accumulates points based on interactions with police officers, non-violent crimes, and parole status. If one collects enough points, they will be added to a list of “Chronic Offenders”. The LAPD will then send officers to monitor that person periodically until they manage to get off the list.

PredPol

Created as a collaboration between the LAPD and UCLA Anthropology professor Jeffrey Brantingham, PredPol is a crime predication software that analyzes historical records to generate estimated times of crime incidents in a particular area around the city. PredPol is mostly used for motor vehicle related crime, specifically burglary and grand theft auto, and is stated as largely a preventative program, seeking to reduce crime rates by increasing police presents in areas predicted to have a crime happen.

Community and LAPD Response

Both of these predictive policing software tools have come under fire from both academics (at UCLA and elsewhere), as well as concerned citizens represented in the SLSC. These groups argue that the presumption of bias-free algorithms is negated by the use of historical crime data as input, and also criticize Professor Brantingham’s role in the collaboration as a violation of anthropology’s ethical obligation to not harm the civilization of study. [21]

The LAPD also published an internal report on the use of predictive policing for the past 10 years. The study found that both the LASER program and PredPol software were ineffective at producing crime prevention and generating information at a rate any greater than the police officers already patrolling the city. The report notes that the racial makeup of both the hot districts and "Chronic Offenders" match the historical data of the 2000s and 2010s well, but this presumes absolute equity in reported crime and arrests for this period as well. [20]

Racial Makeup of the LAPD LASER Chronic Offenders Program (Image: Mark P. Smith, Public domain)

Future of Predictive Policing in the LAPD

As it stands now, in response to the 2019 internal study regarding the inefficacy of current predictive policing algorithms, the LAPD phased out the use of the LASER program in 2019, and PredPol in 2020. [22]Note that this was internally and externally communicated as disappointment with the outcomes of the software and a lack of funds due to the Covid-19 pandemic, and not the underlying principles that some groups are concerned with. A major roadblock is the fact that predictive policing data (and sometimes even use) is not publicly available, and the LAPD has not allowed outside research professionals or LA organizations access.

Predictive policing remains a controversial technology but is still being used in departments across the country. Police departments don't want to get rid of it because they see it as a useful tool to help them perform their job, while the people being policed do not trust the latent functions. In the future, some extensions of this research that may be helpful could be the inclusion of research into how other countries have used predictive policing and the issues they have faced. Additionally, as this page was written during the pandemic, it may be helpful to update how police departments change their usage of predictive policing software after the pandemic as many are currently facing pandemic-related budget cuts, causing them to temporarily abandon the technology.

References

• ↑ a b Pearsall, B. Predictive Policing: The Future of Law Enforcement? National Institute of Justice. https://nij.ojp.gov/topics/articles/predictive-policing-future-law-enforcement.
• ↑ [1], Hvistendahl, M. (2016, September 28). Can ‘predictive policing’ prevent crime before it happens? Science | AAAS. https://www.sciencemag.org/news/2016/09/can-predictive-policing-prevent-crime-it-happens.
• ↑ Bureau of Justice Assistance, & Police Executive Research Forum. (2013). COMPSTAT: Its Origins, Evolution, and Future in Law Enforcement Agencies.
• ↑ a b c Rieland, R. Artificial Intelligence Is Now Used to Predict Crime. But Is It Biased? Smithsonian Magazine. https://www.smithsonianmag.com/innovation/artificial-intelligence-is-now-used-predict-crime-is-it-biased-180968337/
• ↑ Howarth, E. (2018, January 25). Overrepresentation in criminal justice systems. LSE Undergraduate Political Review. https://blogs.lse.ac.uk/lseupr/2018/01/25/overrepresentation-in-criminal-justice-systems/
• ↑ a b c d e f g [2], Brennan Center for Justice (2020, April 1). Predictive Policing Explained. Retrieved December 2, 2020.
• ↑ a b [3], Buzzfeed News (2020, April 21). The Los Angeles Police Department Says It Is Dumping A Controversial Predictive Policing Tool. Retrieved December 4, 2020.
• ↑ a b c d [4], Puente, M. (2019, July 3). LA Times: LAPD pioneered predicting crime with data. Many police don’t think it works. Retrieved November 14, 2020.
• ↑ a b c [5], American Civil Liberties Union (2016, August 31). STATEMENT OF CONCERN ABOUT PREDICTIVE POLICING BY ACLU AND 16 CIVIL RIGHTS PRIVACY, RACIAL JUSTICE, AND TECHNOLOGY ORGANIZATIONS. Retrieved November 14, 2020.
• ↑ a b c d [6], American Civil Liberties Union (2018, March 15). New Orleans Program Offers Lessons In Pitfalls Of Predictive Policing. Retrieved November 15, 2020.
• ↑ [7], Moravec, E.R. (2019, September 5). The Atlantic: Do Algorithms Have a Place in Policing? Retrieved December 6, 2020.
• ↑ [8], Palantir (n.d.). Law Enforcement. Retrieved November 13, 2020.
• ↑ a b [9], PredPol (n.d.). Home page of website. Retrieved November 13, 2020.
• ↑ [10], Big Data, (n.d.). Predictive Policing: From Neighborhoods to Individuals.
• ↑ a b c [11], Collins, D. (2018, July 5). Police1: Police departments sued over predictive policing programs. Retrieved November 14, 2020.
• ↑ [12], PredPol (n.d.). Predictive Policing: Guidance on Where and When to Patrol. Retrieved December 8, 2020.
• ↑ a b [13], Ahmed, M. (2018, May 11). The Intercept: AIDED BY PALANTIR, THE LAPD USES PREDICTIVE POLICING TO MONITOR SPECIFIC PEOPLE AND NEIGHBORHOODS. Retrieved November 14, 2020.
• ↑ [14], Stop LAPD Spying Coalition (n.d.). Website home page. Retrieved December 4, 2020.
• ↑ [15], Baek G., Mooeny, T. (2020, February 23). CBS: LAPD not giving up on data-driven policing, even after scrapping controversial program. Retrieved December 6, 2020.
• ↑ a b [16],Smith, M. P. (2019). Review of Selected Los Angeles Police Department Data-Driven Policing Strategies (pp. 1-51) (United States of America, Los Angeles Police Department). Los Angeles, CA: Office of the Inspector General.
• ↑ [17], Smith, M. P. (2019). Review of Selected Los Angeles Police Department Data-Driven Policing Strategies (pp. 1-51) (United States of America, Los Angeles Police Department). Los Angeles, CA: Office of the Inspector General.
• ↑ [18], Miller, L. (2020, April 21). LAPD will end controversial program that aimed to predict where crimes would occur. Retrieved November 16, 2020

Where this page came from

This page was imported from Wikibooks. From “Lentis” on Wikibooks, by its contributors, under CC BY-SA 4.0. Changed here: set as a page; navigation and edit links left out; each image under its own licence, credited in its caption.

Nobody has written it yet — it is the source material at a new address, which is why search engines are asked to skip it and why no one earns from it. It is up for grabs: take it on, and it is yours to rewrite and to earn from.