Chevron icon It indicates an expandable section or menu, or sometimes previous / next navigation options. HOMEPAGE

Computer algorithms are now deciding whether prisoners get parole

For an increasing number of prisoners, judges aren't deciding whether they get parole. Computer algorithms are — and most inmates don't even know it. 

Advertisement

According to the Wall Street Journal, at least 15 states in the U.S. now use automated risk assessment tools (including software) that consider dozens of factors about a prisoner to aid judges in making parole decisions.

prison handcuffs
Kevork Djansezian / Getty

In the latest episode of Codebreaker, the podcast collaboration between Marketplace Tech and Tech Insider, we talk to experts who think that this could be the future of parole, for better or worse.

Jonathan Levav, a professor of consumer behavioral theory at Stanford, has found that there are distinct patterns in the likelihood of a prisoner getting parole depending on the time of day and whether the judge had just eaten a snack or lunch (Hint: a hungry and tired judge does not bode well for an inmate).

A machine would remove that inconsistency, he tells Codebreaker.

Advertisement

That's one big benefit to having machines help out with — or take over — parole decisions. But algorithms can have questionable moral authority, too. Consider this: what if an algorithm uses data on race or gender to help make a parole decision? Based on past data, a machine might have a significant racial bias. But it's not as if judges are completely race and gender-blind, either.

The factors that go into a parole decision tend to be different depending on whether judges or algorithms are involved. According to the Wall Street Journal, judges tend to consider factors like a crime's severity and the prisoner's level of remorse. The algorithms use biographical data and information gathered in interviews, like a prisoner's age at first arrest, to guess whether they are likely to commit a crime in the future.

Donald Zeller, a former prisoner in Washington, has dealt with algorithms making decisions about his life behind bars. According to Zeller, prison officials told him and his fellow inmates that software would assign the risk classifications that determined what kind of prison — minimum or maximum security, for example — they would go to.

The software considers all sorts of factors, from history of violence and age to the number of felonies and misdemeanors previously committed. 

Advertisement

"It didn't bother us. It speeds up the process. I don’t think it would be different if a person versus a computer decided," says Zeller, who was placed in a minimum-security prison.

He would, however, be much more uncomfortable with a machine deciding whether he deserved parole. "No, I wouldn't like that. A computer can’t calculate an emotional state of mind, how clean a person’s been," Zeller tells Tech Insider.

Still, computer algorithms have been credited with lowering recidivism rates and cutting down on prison populations in states where they are used. That makes them worth seriously considering.


Professor of Criminology and Statistics Richard Berk discusses crime predicting machines with Codebreaker host Ben Johnson. 

Listen to the whole episode for a lot more about decisive machines. Subscribe on iTunes or your favorite podcast app. 

 

Advertisement
Close icon Two crossed lines that form an 'X'. It indicates a way to close an interaction, or dismiss a notification.

Jump to

  1. Main content
  2. Search
  3. Account