Newsgroups: comp.ai,comp.ai.neural-nets,comp.ai.fuzzy,sci.cognitive
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!news.mathworks.com!newsfeed.internetmci.com!psgrain!charnel.ecst.csuchico.edu!csusac!csus.edu!netcom.com!kovsky
From: kovsky@netcom.com (Bob Kovsky)
Subject: Re: AI and Law 
Message-ID: <kovskyDC96sD.LD7@netcom.com>
Organization: NETCOM On-line Communication Services (408 261-4700 guest)
References: <push-2207950118170001@mind.mit.edu> <3v0bcf$je8@cantaloupe.srv.cs.cmu.edu> <kovskyDC8CL4.4Fv@netcom.com> <3v15pa$e3@cantaloupe.srv.cs.cmu.edu>
Date: Tue, 25 Jul 1995 03:53:01 GMT
Lines: 33
Sender: kovsky@netcom3.netcom.com
Xref: glinda.oz.cs.cmu.edu comp.ai:31825 comp.ai.neural-nets:25712 comp.ai.fuzzy:5281 sci.cognitive:8542

	After thinking over my previous article, some additional areas 
of law might be of interest to AI practitioners.

	There are specialized practice areas where a lawyer typically has
over a hundred active files.  None of the cases involves big amounts of
money; usually there are few, if any ambiguities; and each case requires a
more or less standardized course of events utilizing lots of forms.  The
legal issues are generally cut and dried and the lawyer's main task is to
keep everything straightened out and moving forward on time.  He or she
also needs to do some actual negotiating as to a resolution, but that is
almost collateral to the paperwork.  Such areas include workers'
compensation and dissolution of marriage (divorce). 

	Another area of interest would be tax.  Obviously, anyone who
could write any kind of useful program for income taxation would have
earned the Gates-magnitude reward that would be forthcoming; but there are
smaller and more specialized areas of taxation, such as property tax,
where the volume of statutory law to be mastered is much smaller.  The
reason I suggest tax as an area where AI techniques might be interesting
is that it is wholly artificial and the drafters are usually able to write
almost mechanical (but sometimes complex) rules.  Administrative
efficiency and the interests involved encourage rules as nearly mechanical
as possible.  An especially interesting question is whether it is possible
mechanically to identify potential ambiguities, of course without
attempting to resolve them. 

	To follow up on Prof. Fahlman's previous article where he remarks 
that "any number of fields can provide us with similar problems," from my 
experience law is unrivaled in providing explicitly-stated rules 
employing stereotyped language in complex situations where practical 
(commercially valuable) results could be obtained.  It marks a borderland 
between formalisms and common-sense reasonings; and many aspects are 
so highly formalistic as to make them amenable to mechanical approaches.
