Julia Dressel, a student from Dartmouth College, set out to discover whether humans or algorithms were better at assessing risk when determining risk scores for jail sentencing and bail hearings, Science reports. In what is being called the first “horse race” between humans and algorithms, the findings showed that, statistically speaking, humans and machines assessed risk the same. Dressel’s study also revealed, in this instance, both humans and machines were unable to overcome bias. The report notes dozens of states have turned to algorithms to determine court sentencing and bail hearings, with many of them being proprietary and few having undergone independent assessments. 
Full Story