The Compas algorithm, which has worked for the American police since 1998, analyzed the data of the defendants, and then, based on the information received, helped to decide, for example, whether the criminal should be released on bail or should be kept in custody. When choosing a measure of restraint, the system took into account age, gender and track record in the criminal arena. For 20 years of "service" the algorithm was evaluated by more than a million people, but recently it was declared incompetent, after which it was immediately recalled.
Scientists from Dartmouth College tested how accurate the system is and whether it can be trusted. To do this, they recruited freelancers, ordinary people without legal education, to make their own decisions based on short information about people, giving the subjects information about gender, age, criminal history and several other parameters.
The accuracy of the forecast of the freelancers who had a small dossier was almost 70 percent, while the program lagged behind people by five percent, while relying on 137 biography points. Analysis of the judgments of the algorithm showed that black prisoners are more likely to be suspected by the program.
“Mistakes in such cases can be very expensive, so it’s worth considering whether this algorithm should be used at all to deliver court verdicts,” says one of the study authors.
Having studied the principle of the algorithm, the researchers came to the conclusion that the younger the defendant and the more arrests he has behind him, the higher the likelihood of a relapse, therefore, AI specialists recognized the technology as unreliable.
Viacheslav Larionov
