Courts are exploring a new application of machine
learning, using artificial intelligence to predict whether
criminals will re-offend and basing sentencing on those
predictions. At the moment this is being criticized as being
inaccurate, but it is likely that it can become more accurate with
more data. What are the advantages and disadvantages of using
artificial intelligence to make such decisions? How might biases
creep into what seems like an impartial decision because it is
made by a machine? Is it more of a problem because in deep
learning systems we don't know how the computer reached its
conclusion? Who benefits and who is harmed by such a system?
Should we improve and use this technology or is there a convincing
moral argument against it?