clock menu more-arrow no yes

Filed under:

Can the algorithm police use to predict crimes be racist?

Police departments around the country are using data and algorithms to predict which criminals are likely to commit certain crimes, but studies show those programs could disproportionately target minorities.

Known as "predictive policing," police officers can use various technologies to analyze prior criminal history and find patterns to help predict and prevent future crimes. Data can include demographic information, parolee populations, economic conditions, prior crime trends and environmental data, according to the National Institute of Justice.

But when ProPublica investigated one such program in May, it found that a program used by police in Coral Springs, Florida, routinely rated black criminals at a higher rate of committing a future crime than white criminals.

The investigation found that only 20 percent of those who predictive policing measures decided were most likely to commit violent crimes in that county actually did. And when they factored in all crimes, ProPublica reported the algorithm was only "somewhat more accurate than the flip of a coin," with 61 percent of those predicted to commit another crime went actually going on to do so.

"The formula was particularly likely to falsely flag black defendants as future criminal, wrongly labeling them this way at almost twice the rate as white defendants," the investigation reported.

It went on to note that white defendants were "mislabeled as low-risk more often than black defendants."

The Chicago Police Department also used predictive policing technologies called the Strategic Suspects List.

After learning that the list identified "under 1 percent of homicide victims for that year," the department revamped the algorithm and improved that number to 29 percent.

There has been evidence that predictive policing could be helpful for departments, including a 2013 study from the University of Pennsylvania.

The study also goes on to explain the complexities of predictive policing, saying that predictive policing technologies can "forecast more accurately" than conventional human predictions, but that it depends on the complexity of the matter at hand.

"In criminal justice settings where real lives can be at stake, the consequences could be significant. Why take the risk?" the study stated.

Another study found predictive technologies "can disrupt opportunities for crime and lead to real crime reductions," though it's worth noting it was co-authored by developers of predictive policing technologies.

Email:; Twitter: @sarapweber