preview

Racial Bias And Momendations In The Criminal Justice System

Decent Essays
Open Document

This overall inability to predict crime is due to many discrepancies found in the risk assessment. First white defendants were predicted to be less risky than they actually were with white defendants mislabeled low risk almost twice as often as black defendants. Black defendants were 45% more likely to be assigned higher risk scores than white defendants when controlling for prior crimes, future recidivism, age and gender. Black defendants were also twice as likely to be misclassified as being higher risk for violent crimes than white defendants while white defendants that do reoffend with violent crime were 63% more likely to have been misclassified as low risk for violent crime. Finally, black defendants were 77% more likely to be …show more content…

Additionally, data from internet will not be a good way to determine whether a person is a terrorist especially if that person has shown themselves critical of the Trump administration. This algorithm may classify them as terrorist simply because they don’t agree with current policies while clearly disagreeing on policy does not many a person a terrorist. There are many approaches and algorithms involved in machine learning. Some such algorithms are decision tree learning, association rule learning, reinforcement learning and rule-based machine learning. All these algorithms sound similar to what the names suggest in which decision trees are made to make decisions and learn, association rule learning is from discovering relations between variables, reinforcement learning is when it learns what to do through reinforcement, and rule- based learning is where rules are stored about how to use knowledge. While these all can mirror to some extent variations of human learning and decision making, not every aspect that occurs in a human brain is in the algorithms that exist such as the higher-level processes that occur within decision making and learning so therefore it is difficult to mirror all human aspects of learning and behavior in AI. Due to these differences in learning, prejudice in AI is different than the usual mechanism seen in people. While AI is able to learn stereotypes, and pick up on language bias like humans do through types of

Get Access
Related
Get Access