At the recent Legal Futures conference (London, November 19^th 2012) Daniel Katz, Assistant Professor of Law at Michigan University suggested that there is something of a renaissance going on in AI (Artificial Intelligence) and Law.  Professor Katz gave a really entertaining talk where he explained about Quantitative Legal Prediction.  Essentially this is where very large data sets are interrogated from which computer software appears to learn and can thereafter predict outcomes.  Two obvious application areas for this technology would appear to be E-discovery and analysis / prediction of judicial decision making.  The rationale for this technology was said to be that human experts, although very good at pattern recognition, simply cannot absorb and traverse the huge amount of data which is now becoming increasingly available to all thanks to the internet.  There is much more information available at Professor Katz’s Computational Legal Studies Website.

Back in the late 1980’s I completed an MSc in AI (the course was actually named ‘Knowledge Based Systems’) and I was, surprise, surprise, the only lawyer on the course at that time.  It paid dividends though as I got to work with Richard Susskind as a result, just after the appearance of the Latent Damage Act system (LDA) – the world’s first commercially available legal expert system.  Expert systems are a branch of AI and they are generally considered to be more rule based systems.  So far as I know there had been no other legal expert system on the market until very recently.  The Road Traffic Representation System (RTRS), developed by Martin Langan, appears to be a mix of a rule based expert system and a legal workflow system.  Martin also gave a presentation at the Legal Futures conference and he explained that users can login, provide answers to a series of questions about their road traffic accident and the system will then offer some legal advice – for free.  Users can also, if they wish, be put in direct contact with a lawyer and have a telephone conversation (for a fixed fee) and / or engage a barrister to represent them in court.

Expert systems like the RTRS and LDA are designed by people known as knowledge engineers.  Such people may or may not do software coding.  Their main role is to understand how experts (in the present context, legal experts) solve problems, and then try to represent that problem solving skill in a way which can be coded in computer software.  An obvious starting point for this is to devise check-lists which mimics the experts thought process and which may be composed of questions the experts will themselves ask clients.  Generally however, when processing large amounts of complex data, experts will have their own heuristic rules – derived from their knowledge and experience – rather than work by detailed checklists.  Arguably, identifying and representing these heuristic rules is the ultimate goal of the knowledge engineer.  As you might expect, it helps enormously if knowledge engineers have themselves some knowledge of the subject domain.  It is no co-incidence therefore that Martin Langan is a lawyer (and used to be a partner in a law firm) as is of course Richard Susskind (the legal expert of the LDA system was Professor Philip Capper).

Legal expert systems seemed to have reached something of a cul-de-sac some years ago.  Although a lot of interesting legal AI research has continued to be conducted for many years, for practical purposes legal expert systems seemed to have been overshadowed by more directly relevant (and lower-tech) tools such as litigation support databases and hypertext systems made up of related linkages.  If there is indeed a renaissance of legal AI (including expert systems) then this should be good news for many law graduates.  Legal knowledge engineers and system developers will become increasingly in demand.  Many practising lawyers should also welcome such systems.  The LDA for example was designed to be used by lawyers who were themselves not expert in the intricacies of the Latent Damage Act, but who needed to make an assessment about whether a case concerning latent damage may or may not have been barred by the limitation acts.  With the RTRS too, I wonder how many motorists who, because they are made more aware about their legal circumstances via the system, then decide to instruct a barrister when they otherwise might not have?  Arguably, the more that non-lawyers know about the law and their legal rights translates directly into increased demand for lawyers’ services.

It is possible to approach AI and law by examining some very fundamental philosophical issues (such as, for example: what is knowledge?  What is legal knowledge?  Can legal knowledge be truly represented or captured in a software application?) and also more prosaically, with emphasis placed on practical problem solving.  I for one welcome an AI renaissance and I am sure that, given the interests of the LawSync team, undergraduate students taking the LawSync module at SHU will be introduced to some aspects of this fascinating area which looks set to become increasingly important as the legal services market develops.

(Image from Chris Devers’ photostream http://www.flickr.com/photos/cdevers/4456489056/ Kismet the AI robot at the MIT Museum)