Inleiding en context
Goede morgen allemaal op deze dinsdag 11 januari 2022. Vandaag een ´kennisparel´ over het gebruik van zogenaamde ´big data´ door de Nederlandse politie en de effecten hiervan op die organisatie. Het gebruik van ´big data´ wordt door de politie veelal gebruikt om verschillende misdaadpatronen in tijd en ruimte te karakteriseren en om deze kennis te benutten voor de preventie en opsporing daarvan. Er wordt ook wel gesproken van proactieve interventie-inspanningen. Er zijn verschillende voorbeelden van voorspellende politieprogramma’ waar voorspellende analyses de politie in staat stellen om potentiële schade aan kwetsbare gemeenschappen tot een minimum te beperken en kunnen zorgen voor een evenredige verdeling van de voordelen van bijvoorbeeld criminaliteitspreventie onder de bevolking en binnen geografische gebieden.
Big data, het is bijna een toverwoord, maar de auteurs van bijgesloten ´kennisparel´ maken een groot aantal relativerende opmerkingen over nut en onnut van de toepassing hiervan. Kortom: opnieuw een voorbeeld van ´ontmythologisering´ van denkbeelden en veronderstellingen over grote thema´s, waaronder ´big data´, binnen het domein van criminaliteit en rechtshandhaving. En om het af te leren, een ´gouden oude´ uit de rijke historie van de Nederlandse popmuziek: https://www.youtube.com/watch?v=fVpv5MnVoE0 Prachtig.
Bron
Schuilenburg, Marc & Melvin Soudijn (december 2021). Big data in het veiligheidsdomein: Onderzoek naar big data-toepassingen bij de Nederlandse politie en de positieve effecten hiervan voor de politieorganisatie. Tijdschrift voor Veiligheid, vol. 20, no. 4, pp. 1-19.
Summary
In recent years, big data has revolutionised many domains, including policing. However, there is an insufficient insight into the different applications used by the police and the potential benefits of big data analysis for the police. Literature on big data and policing mainly focuses on predictive policing and the associated risks. This article provides an up-to-date overview of the various big data applications by the National Police in the Netherlands. We distinguish three areas: work on the street, investigation and investigation, and intelligence. The research shows that hardly any big data applications work fully autonomously (without human intervention). For the time being, it concerns fairly simple algorithms and big data applications. Finally, we distinguish two positive effects of big data applications for the police organisation: accelerated learning and a single police partnership.
Big data applications are becoming an increasingly important part of the functioning and actions of the Dutch police. Important reasons for this are the increased technical possibilities for collecting, processing, and analyzing large amounts of data and the possibilities for using the results in police work. In addition, political and cultural developments play an important role in this, from the political will to tackle certain forms of crime at an increasingly early stage to a strong belief in technology as ‘the’ solution to social problems such as insecurity. Despite the growing popularity of and attention for big data, there is a lot of uncertainty about its application and effects among the police. Not only is the big data area surrounded by all kinds of technical terms that are often used interchangeably, confused with each other or about which there is little or no consensus about the exact meaning. The goals and time perspectives of big data applications can also differ drastically. Moreover, the application of big data within the police sector covers a broad field, from the targeted collection of data for crime detection to the automation of internal company processes.
Based on the research, it can be concluded that big data applications are used very widely by the police. For example, traditional enforcement on the street and investigation and intelligence systematically use numerous applications and possibilities in the field of big data. It is striking that the application of big data for forms of risk assessment is still the least common. In fact, it often concerns applications with which large data files are linked, so that information can be found more quickly. The second observation is that the broad application of big data by the police has several possible positive effects. These have been broadly termed ‘accelerated learning’ and ‘a police partnership’ based on literature research. This means that the police force can ‘learn faster’ by processing large volumes of data via algorithms, especially when a feedback loop is created, whereby new data generate different outcomes or can be reacted differently by different parts of the police organization. In addition, and this is closely related to the previous effect, allowing the police organization to move more homogeneously, with clear priorities and more targeted management.
In interpreting the application of big data in the police organization, the literature speaks of a ‘second revolution’, ‘a new era’ and a ‘paradigm shift’. The research shows that for the time being, relatively simple big data applications are involved, such as detection apps with fairly simple algorithms for working on the street or linking and creating large data files. This type of use is also jokingly referred to as ‘little data’, indicating that an algorithm is not completely given free rein but is approached through certain hypotheses. In other words, these are forms of feeble artificial intelligence. As a result, the impact for citizens, such as discrimination and ethnic profiling, appears to be limited for the time being.
However, it is expected that more complex – fully self-learning – algorithms will be used to combat and prevent crime in the future. It is known from the literature that potential risks such as discrimination and ethnic profiling can play a greater role than with simple -rule-based algorithms . After all, the application will become increasingly complex because the choices are not programmed completely freely but are based on the data and experiences of the application itself. Ethical principles, as mentioned by the High-Level Expert Group on Artificial Intelligence of the European Commission in Ethics Guidelines for Trustworthy AI, as ‘respect for human autonomy’, ‘prevention of harm’, ‘fairness’ and ‘explicability’ are therefore becoming increasingly important for the police organization when developing new big data applications. According to ethics by design, the basic principle here is that relevant public values are protected by making them known at the earliest possible stage of the design process and incorporating them into the technology.
It is interesting for follow-up research to see whether and how the police organization will be able to integrate ethical values in the design of (new) big data applications. In any case, the first signs are favourable. The police recruitment sites indicate that the police intend to work as much as possible with open source technology for solutions involving big data. The results (meaning the apps and algorithms) would then be shared in the public domain. This idea is that such applications should not become a black box but should be open and transparent. It is also noted that it is important to think about these issues. Otherwise, the enormous volume and variety of data, especially from digital seizures, can overwhelm the police.
Afsluitend
In situaties waar fysieke vrijheid of persoonlijke veiligheid op het spel staan, zoals bij predictive policing, de bepaling van recidive risico´s en veroordeling is het zaak om hier zeer kritisch op te zijn. Wanneer een big data bijvoorbeeld wordt gebruikt voor opsporing of veroordeling kan dat vertrekkende gevolgen hebben voor de betrokken mensen. Het kan fungeren als een black box waardoor het onmogelijk is voor juridische professionals, zoals rechters, advocaten en officieren van justitie om de redenering achter de uitkomsten van het systeem te begrijpen. De toekomst van het toepassen van voorspellen en risicobeoordeling binnen de politieorganisatie is dichtbij en eigenlijk al gearriveerd. Critici wijzen op de noodzaak van transparantie en toezicht op de door de politie gehanteerde nieuwe methoden. Ondanks goede intenties kunnen deze onder meer discriminatoire prakijken bevorderen en een bedreiging vormen voor elementaire burgerrechten. Ten slotte: ´bigger is not always better´, er kan inderdaad sprake zijn van een overweldigende hoeveelheid data waardoor het volgende effect kan optreden dat ook wel bekend staat onder het fenomeen van ´paralysis by analysis´. En dat kan natuurlijk ook niet de bedoeling zijn.