Electronic judges; and decision support tools

Back

What technology cannot do is say "that witness is a liar". IT can help make that decision.

IT can diagnose that a particular sentence is about right, having regard to the circumstances of the offender, the circumstances of the offence and the effect of these on each of the sentencing purposes, but it can neither explain the outcome nor provide a watertight diagnosis when the case is one outside the usual pattern. We have not even begun to take advantage of IT in sentencing. We are using it to see if a particular penalty is disparate, but no-one, so far as I know is using, for example, neural nets to find out what penalty might be proportionate.

We need to know what penalty might be "about right" in the particular case before we can ask "is it disparate with the penalty others are receiving in other court rooms, or in my court room yesterday?"

IT cannot identify the basic sentencing principles, such as proportionality, disparity, totality. It can bring the fact they exist to people who have no choice but to do their best to impose a fair and just sentence. It can help identify what is fair in the particular case and what will be just having regard to the cases that have gone before and the cases that will follow, but it cannot strike that balance.

If we seek to compete with IT on its own ground, we will fail, and fail miserably. At first instance, there is a role for the electronic judge, of the type we already see enforcing traffic laws and of other kinds, which can help people resolve their grievances. If no-one else has them, come to South Australia and we will show you ours.

There is another role when teh judge is electronic, a role with its roots in the doctrines of natural justice: the role of reviewing and controlling what the electronic judge is doing. The electronic judge can outline the case against an individual. The electronic judge cannot know whether it is outlining that case to the right person. The electronic judge cannot provide the solution when the individual wishes to challenge the case against them.

Other people in the community have reason to fear the electronic judge, but we and they will have no reason to be frightened of the electronic judge if we can keep him confined to the task he is good at.

We can use the electronic judge as a sifter and as a mediator in the cases so factually complex that they are beyond the capacity of one person to absorb, as the parties did before they went to litigation about a huge dispute which arose out of the European Chunnel project, following a procedure which has been described in an abstract stored online in one of the decision analysis archives.

There, an electronic judge was used to find and explain the strengths and weaknesses inherent in each case, and to assess the risks associated with litigation, using neural net technology developed in the course of the search for artificial intelligence. Once this occurred, it became clear that the real disputes were disputes the parties could resolve between themselves without going to a 'real' judge and drowning him in years of litigation and filling buildings (or computer hard disks) with exhibit documents.

We know what it is hard for us to do. We know what we do well. We need to say what these things are, in terms that IT people can grasp, and to look for ways of using IT to do the things we cannot do well, to help us do what we do do well.