Artificial Intelligence Previews Alternative Final Judgments

I have a different take from the author’s on the interesting article from Australia which appears below. Artificial intelligence (‘AI’) is impressive (although it still has a way to go).

But more to the point, all software (even software that doesn’t claim to be AI) functions according to rules and formulas built into it. Unless and until the underlying process intended to be automated can be boiled down and captured in detailed, hard-and-fast rules and/or formulas, no software can do (or be expected to do) the job. Period.

That’s not the limitation or the fault of the computers or the software.

Rather, that’s because the underlying process the programmers are attempting to automate or computerize is more art than science, sometimes hinging, in the final analysis and in truth, on impressions and gut-feelings for the particular situation and people.

Because of the importance in many cases of that je ne sais quoi element, the artist-practitioners of the process (legislators and judges) haven’t yet been able (and probably never will be able) to nail the entire dissolution process down to tight, iron-clad rules and formulas that can be handed off to the programmers.

The large number of je ne sais quoi cases and the je ne sais quoi elements of many other cases is why human judges don’t have to worry about being replaced by computers, and why lawyers will not become ‘unnecessary’ or dispensable in the divorce process any time in the foreseeable future.

Family court is not tax court. Discretion and judgment are vital to doing the right thing for the ever-growing numbers of regular people drawn into the legal system to end, modify, regulate and get supervision of their family relationships (or aspects of them).

Lawyers and judges earn most of their keep working very hard in the trenches of je ne sais quoi-land:

  • the grey-shaded fringe territories of child support, alimony and property division and
  • the vast formula-and-hard-and-fast-rules-defying precious regions of child custody and visitation.

In these cases especially, people need vigorous legal representation (not just advice) and a Solomon-wise judge. Neither AI or ordinary computer software is going to change that anytime soon, and probably never.

I’m an avid booster of technology – for those tasks for which it is well-suited. It’s not well-suited for the cases bulleted above.

(No, child custody should not be awarded to a parent simply because that parent wants it the most and allocates the most points to it. Any child-molesting parent out there would likely win custody with that approach. )

Legislators and judges have gotten extremely good at “scientificizing” much of child support calculations and property division and, in some jurisdictions, to a lesser extent, alimony calculations.

For those cases where disputes are confined within the black-and-white boundaries of such subject matter, the AI software described in the article may help amicable spouses visualize realistic alternative monetary outcomes and thus motivate and encourage them to reach a settlement, rapidly, easily and inexpensively.

As long as such software’s capabilities and limitations are fully understood by its users, the legal system should embrace its use in appropriate cases. That just may free up scarce legal system resources for the difficult cases where lawyers and judges are needed the most and contribute the most.

Divorce? Let the computer be the judge

By Adele Horin

September 21, 2005

She wants the cat and so does he. He wants the car and so does she. And they both want the kids.

After a marriage breaks down, couples can spend a fortune in legal fees wrangling over property and custody.

But a new computer program called Family Winner may short-circuit the court battles. Developed by Emilia Bellucci, a lecturer at Victoria University’s school of information systems, and John Zeleznikow, professor of information systems, the program requires a couple to prioritise their demands, assigning each a numeric value so that the sum is 100.

“Then the program distributes everything according to who wants it the most,” Professor Zeleznikow said.

The program is part of a package he hopes will be taken up by the 65 Family Relationship Centres being funded by the Federal Government.

With separating couples required to try to resolve disputes before filing with the Family Court, Professor Zeleznikow expects there will be a need for mediators and counsellors to deliver information quickly and cheaply. Computer programs that show the likely outcomes of court cases, or apply artificial intelligence and game theory to the allocation of possessions, may be part of the answer, he believes.

A second program called SplitUp, developed with Andrew Stranieri of the University of Ballarat, calculates the likely results of property settlements in Family Court proceedings, Professor Zeleznikow said.

The program had highlighted surprising results, he said, such as the importance judges assigned to the relative ages of the protagonists, and the lesser significance attached to the assets each brought into the marriage.

“I was surprised to see that if a woman with little work experience is, say, 10 years younger than her husband who is a professor on a good salary, the assets are likely to be divided reasonably equally, when you might expect the professor would need less,” Professor Zeleznikow said.

“Are these systems better than a good family lawyer? Absolutely not. But you will pay a lot of money and take some time to get advice from a good lawyer. And this could be available for everyone.”

For informational purposes only and not intended to infringe on Copyright © 2005 The Sydney Morning Herald