Police departments in several large U.S. cities are experimenting with predictive policing, utilizing computer systems to analyze extensive datasets that include historical crime data.
The analysis aids in determining optimal police deployment locations and identifying individuals deemed at higher risk of either committing crimes or being victims themselves, according to the Brennan Center for Justice.
In a workshop with The National Academies of Sciences, Engineering, and Medicine, panelists discussed the ethical implications and practical limitations of predictive policing in criminal justice.
Speakers emphasized the need for careful consideration in implementing predictive technologies, highlighting potential risks such as bias, accountability, and the replacement of human discretion with algorithms.
They also stressed the importance of addressing broader social and economic factors contributing to crime in marginalized communities and evaluating predictive policing technologies to balance civil liberties and public safety.
Renée Cummings, artificial intelligence (AI), data, and tech ethicist, and the first Data Activist-in-Residence at the University of Virginia’s School of Data Science, believes that while predictive policing can help in certain aspects of crime fighting, it’s not a tool fully developed for deployment.
“What we are seeing at the moment with algorithms, not only in housing, and health care and education and finance is that we still don't understand how we are coding risks, how we are coding what has seemed to be a violent or nonviolent criminal or not criminal,” Cummings said.
Elizabeth Joh, the Martin Luther King Jr. Professor of Law at University of California Davis, believes predictive technologies theoretically offer substantial potential benefits because previous policing relied heavily on human decision-making, which involved significant discretion and subjective judgments.
However, Joh says the technique should not be a tool to replace human decision-making. “I think the problem here is that instead of thinking about prediction as an aid to human decision making, we're seeing many instances in which the predictive technology is a substitute for human decision making,” she said.
The Los Angeles Police Department (LAPD) began experimenting with predictive policing alongside federal agencies in 2008. The department has since launched programs like LASER, targeting potential gun violence areas, and PredPol, identifying high-risk spots for property crimes, according to the Brennan Center for Justice.
In 2017, The New York Police Department (NYPD) developed predictive algorithms for various crime types such as shootings, burglaries, felony assaults, grand larcenies, motor vehicle thefts, and robberies. The algorithms assisted in deploying officers to monitor specific areas effectively.
The Chicago Police Department also launched a major person-based predictive policing initiative, known as the "heat list" or "strategic subjects list," initially piloted in 2012. This program identifies individuals deemed at high risk of committing or being victims of gun violence, the Brennan Center for Justice said.
One important aspect of the conversation about predictive policing is how it affects community trust and faith in the criminal justice system.
Thomas Abt, Founding Director of the Center for the Study and Practice of Violence Reduction at the University of Maryland, believes that to encourage trust from the public, the criminal justice system must adhere to fundamental principles of procedural justice: being inclusive by not just including but giving voice to those involved, and establishing an impartial and transparent process.
Former Police Chief Jim Bueermann, founder of the Future Policing Institute, says the process of evaluation involves understanding and addressing community concerns, particularly in high-crime neighborhoods. “I think is fundamental to enhancing or keeping whatever level of trust and confidence people have in the police,” he said.
The program last week wrapped up a two-day public workshop. The National Academies will publish its findings.