1

Introduction

On June 24 and 25, 2024, the National Academies of Sciences, Engineering, and Medicine (National Academies) held a two-day public workshop exploring law enforcement’s use of person-based and place-based predictive policing strategies. The workshop was a direct response to Executive Order No. 14074 (2022), on enhancing public trust and safety through accountable policing and criminal justice practices, and Executive Order No. 14110 (2023), which addresses the use of artificial intelligence (AI) in law enforcement. Among other measures, the executive orders called for a National Academies workshop to explore the use of predictive algorithms by law enforcement. A committee was charged with developing a workshop to address law enforcement use of person-based and place-based predictive policing approaches, and to explore key considerations for implementation around their reliability, legal implications, disparate impact concerns, and their intersections with privacy, civil rights, and civil liberties. The workshop was planned according to a Statement of Task (Box 1-1) and was framed around several questions related to predictive policing strategies.

David Weisburd, George Mason University and chair of the workshop planning committee, began the event by noting that the workshop was organized in response to executive orders elevating matters of public concern, including advancing effective, accountable policing and enhancing public trust and safety. Weisburd noted specific public concern surrounding the idea of predictive policing, as well as critiques of specific implementations of predictive policing. In developing the workshop, he explained, the committee discussed several challenging questions surrounding predictive

policing approaches. First, there is a lack of clear, shared definitions of predictive policing (Box 1-2). A related challenge, Weisburd noted, was that the term “predictive policing” has largely fallen out of use by law enforcement agencies, which may be due in part to public critiques of certain applications of predictive policing. Predictive technologies include “automated,” “dynamic,” or “data-driven” approaches; however, predictive policing is generally seen as involving the use of algorithms that bring together such technologies to define specific targets and times for police intervention. Regardless of the term used, law enforcement agencies continue to use data collection and analysis tools to anticipate crime and facilitate police response. The needs that predictive policing aims to address, including patrol allocation, remain real-world challenges for law enforcement officials, said Weisburd.

Weisburd noted that the workshop was taking place at a crucial juncture for predictive policing technologies. He explained that while early attempts at predictive policing have largely fallen out of favor, the underlying technologies—such as algorithmic analysis and big data—continue to evolve and find new applications in law enforcement. This progress continues despite the lack of a clear, shared definition of predictive policing. “We may

be on the precipice of a new era of predictive policing,” he said, “with the time and wisdom to consider what that could and should look like.”

Weisburd introduced Nancy La Vigne, National Institute of Justice (NIJ), who provided a brief overview of NIJ’s perspective on predictive policing. She began by briefly recounting NIJ’s history of investing in aspects of predictive policing technologies, and she acknowledged the unintended outcomes of some of these investments. The intent of NIJ’s investments in predictive algorithms was to identify future crime concentrations, with the goal of discerning the factors or correlates that may represent the root causes of crime hot spots so those factors could be addressed. However, in practice, predictive algorithms have fueled hot-spot policing that too often results in excessive policing of communities and residents, imposing biases that have detrimental impacts on people of color, she stated. To mitigate this trend, La Vigne emphasized the importance of focusing on police actions

taken to address predicted hot spots. She explained that the field would benefit from improved guidance on that front, to ensure that police do not simply engage in increased enforcement activities at the expense of both community trust and public safety. She cautioned that even if predictive algorithms were no longer used, police would continue to use data to guide decisions about resource allocation, which underscores the importance of prescribing the types of actions police take in response to predictions. “Predictive policing is not just about the prediction but about the type of policing that happens after the prediction,” said La Vigne. Finally, La Vigne emphasized that the workshop’s conversation is ultimately about public safety, and she encouraged consideration of the criminal legal system’s responsibility toward communities experiencing high rates of crime and how those communities might be best served.

These opening remarks set the tone for the workshop, emphasizing the importance of addressing the challenges and critiques of predictive policing while exploring its potential benefits and implications for law enforcement, public safety, and community trust.

WHAT IS PREDICTIVE POLICING?

Thomas Abt, Center for the Study and Practice of Violence Reduction at the University of Maryland, offered a concise overview of predictive policing. In a 2009 solicitation, NIJ defined predictive policing as including “strategies, tactics, and enabling tools and technologies that improve the situational awareness of law enforcement concerning individuals or locations before criminal activity takes place.”¹ Abt noted the significant breadth of this definition, echoing Weisburd’s earlier remarks on the difficulty of drawing clear boundaries for predictive policing.

In practice, said Abt, predictive policing programs often use large and varied data sets; employ advanced analytic techniques utilizing algorithms, machine learning, and/or AI; and make predictions or forecasts of criminal behavior. Predictive policing programs are used for law enforcement purposes (although other decision makers and service providers may be involved in their use) and are often developed by commercial vendors. Most predictive policing focuses on predicting places where crime is likely to happen (i.e., place-based predictive policing), but efforts have also been made to use these tools to identify individuals who may be victims or perpetrators of future crime (i.e., person-based predictive policing). (See Box 1-2 for other definitions, and the introductions to Chapters 3 and 4 for expanded definitions and examples of place-based and person-based predictive policing.)

___________________

¹ https://nij.ojp.gov/sites/g/files/xyckuh171/files/media/document/NIJ-2009-2239.pdf

Throughout the workshop, several speakers encouraged greater attention to the “policing” element of predictive policing. Speakers emphasized that predictive policing does not just encompass prediction but also law enforcement activity in response to predictions—and that the nature of these responses is essential to central questions of reliability, disparate impact, civil rights and liberties, and community trust.

ORGANIZATION OF THE WORKSHOP

The workshop was divided into six sessions, focusing on place-based and person-based predictive policing, theoretical underpinnings, examples of use, evidence of effectiveness, and key legal, ethical, and social considerations. In addition, a roundtable conversation with community representatives explored community concerns with and responses to predictive policing. The final session provided an opportunity for the planning committee, speakers, and audience members to reflect on key takeaways from workshop discussions. This proceedings is organized into six chapters to maximize clarity for a broad audience and does not directly follow the event’s agenda order. Following this introductory chapter, Chapters 2 and 3 examine place-based and person-based predictive policing, respectively, exploring law enforcement use and evidence of effectiveness. Chapter 4 discusses overarching legal, ethical, and social considerations for law enforcement use of predictive policing technologies. Chapter 5 addresses community responses to predictive policing, and Chapter 6 summarizes discussions about the future of predictive policing and key messages from the workshop. The workshop rapporteur has prepared this proceedings as a factual summary of the workshop. The planning committee’s role was limited to planning and convening the workshop. The views contained in the proceedings are those of individual workshop participants and do not necessarily represent the views of all workshop participants, the planning committee, or the National Academies.

This page intentionally left blank.