Experiments in US courtrooms suggest that common risk assessment algorithms make harsher recommendations than human judges. This is likely because they are less able than humans to predict which defendants will violate pretrial agreements.
“Some jurisdictions wanted to work with us to assess whether these recommendations are actually helping judges make better decisions.” Kosuke Imai at Harvard University.
In the U.S. criminal justice system, judges decide whether defendants should stay at home awaiting trial or at home.
Source: www.newscientist.com