It may sound like the plot of the 2002 film Minority Report, but artificial intelligence can predict any arrest within three years of a prisoner’s release on parole.
Machine learning was developed to determine a prisoner’s risk of early release by analyzing 91 variables including age, race and previous arrests.
Researchers at the University of California, Davis (UC Davis) used data from more than 19,000 inmates scheduled for the New York State Parole Board from 2012-2015.
Court documents show that 4,168 people were released, but that AI found the board could have released double the number of inmates without increasing the subsequent arrest rate.
Machine learning was developed to assess a prisoner’s risk of early release by analyzing 91 variables including age, race and previous arrests
The AI mirrors the storyline of Minority Report, which starred Tom Cruise.
Set in the year 2054, the film is about a specialized police department that arrests criminals using the foreknowledge of three psychics known as the “Precogs”.
The machine learning algorithm found that the predicted risks for those who were denied parole and those who were paroled are very similar.
This suggests that low-risk individuals may have remained incarcerated while high-risk individuals were released.
Hannah Laqueur, an assistant professor in the Department of Emergency Medicine and lead author of the study, said in a statement, “We cautiously believe the board could more than double the discharge rate without increasing the overall or violent arrest rate.
“And they could have made those gains while eliminating racial disparities in release rates.”
From 2012 to 2015, a total of 19,713 people had parole hearings.
During those years, 4,561 people were granted parole and 16,068 were denied one or more times.
In addition to age and race, the AI also looked at specific offenses such as underage, drugs, hate crimes and theft.
And the AI was fed information about the arrest history.
These included offenses that resulted in the length of incarceration for which individuals are being considered for parole.
The AI found that the number of inmates paroled should be nearly 50 percent higher.
“Our findings are consistent with the Vera Institute of Justice report finding that the board denies parole to many individuals with low-risk COMPAS results and the general complaint that the board is driven by retaliatory impulses, risk aversion and fear of the political Impacts they might face if, for example, they released a person convicted of murder who was then arrested again for murder,” says the study, published in the Journal of Quantitative Criminology.
“This suggests the board is either seeking retaliation or guarding against political backlash.”
The authors note that they do not advocate replacing human decision-makers with algorithms to judge who should be released from prison. Instead, they see a role for algorithms to diagnose problems in the current parole system.
Minority Report, starring Tom Cruise, is about a specialist police department that arrests criminals using the foreknowledge of three psychics known as the “Precogs” (pictured).
“This study demonstrates the usefulness of algorithms for evaluating criminal justice decision-making,” said Laqueur.
“Our analysis suggests that many individuals are denied parole and jailed beyond their minimum sentence despite posing a low risk to public safety. We hope that by providing data on predicted risks we can support reform efforts.’
If you liked this article…
AI detective Valcri solves crimes by making deductions from MILLIONS of pieces of data
AI can successfully identify the likelihood of domestic violence 90% of the time
AI that examines CCTV to predict violent crime BEFORE it happens is being rolled out in the UK