Monday, May 4, 2015

Dear Watson - I have a question for you! Watson & Legal Research: We know it can answer, but can Watson ASK questions?

My colleague Ron Friedmann  @ronfriedmann  posted an interesting question on his blog Prism Legal today. He is collecting "crowd sourced" questions  which he can ask at this week’s IBM’s World of Watson conference. I am fully booked, so I can’t attend, but I have a question for Ron to ask on my behalf:   “Can Watson ask questions?” 

Answers are Easy--- Questions are The Sign of a Pro!

The test of Watson as a legal researcher will not be whether it  can provide a generic legal answer but whether it can ask the necessary series of narrowing questions to help a lawyer define the answer that  she/he really needs.

Watson Beats Humans on Jeopardy - copyright IBM
I am puzzled by the rather giddy certainty among the “legal techno pundits” suggesting  that since Watson has made such great strides in responding to medical questions from doctors,  that it will soon be the ultimate  24/7 associate----Spewing answers at midnight and never uttering a word about  work-life balance. Will it really be that easy to teach Watson to conduct legal research in in the next decade?

Medical Research vs Legal Research

Although I claim no expertise in medical research, I know enough about the evolution of online medical research systems to suspect that there is at least one significant difference between medical and legal research. Medical information has the benefit of having a comparatively standard and nearly universal taxonomy. Symptoms, diseases, diagnoses, adverse reactions are the same in each state and in each country. Measles are the same in New York and California.

 But the law???? There or is no standard naming convention across the federal government and the 50 plus jurisdictions of the United States. Terminology and even the definitions  of ordinary words such as "person" or "homicide"  could differ wildly when you cross state lines. Then there are such unruly concepts as elements, defenses, statutes of limitations, jurisdictional and procedural issues. And yet we are not done.  The federal government and each of the fifty states have not only enacted laws but they each have courts which interpret the application of laws.Then there are the administrative agencies which generate regulations for the federal government and each state - which explain how to comply with those laws. These agencies may have quasi- judicial enforcement arms which generate even more interpretive materials. I am not even going to mention, municipal laws, county zoning, equity, conflict of laws,international treaties, SRO's (self regulatory organizations), the ABA or state bar ethics rules.... and I trust my readers will come up with a dozen more sources of laws and compliance which lawyers need to take into consideration.
Comparing 50 state Laws 
Let's look at the existing state of legal taxonomny and the standardization of legal concepts across the US.  West/Thomson Reuters developed the most  sophisticated taxonomy and normalizing framework for US legal research in its topic and key number system which is compiled in the "Analysis of American Law."  Even its detractors have to admit that it is the closest thing we have to a  taxonomy of US law - and it took over 100 years to develop!
Yet lawyers still struggle to find new tools which modulate and standardize the analysis of laws across the 50 states. 

Compiling a "50 state survey" on a single issue used to be a surefire summer project which might take an associate a whole summer to complete! Thankfully the major legal publishers have spent the past decade trying to tame this particular beast. And today... if an associate gets such an assignment and has access to a savvy  research professional - they may find that a comparison chart, survey or "smartchart" on their issue can be generated in a matter of seconds using one of the premium legal research services (LexisNexis, Blaw, Westlaw, Wolters Kluwer.) But the development of these charts involved a lot of "heavy lifting" by each publisher. There are still  thousands of legal issues  which have not yet been tackled by any of the majors.
It is in fact, a day for celebration when one of major legal publishers releases a new topical survey or a new tool  for comparing 50 state laws which normalizes and highlights the differences and commonalities of laws on a single issue across the country.  

Can Watson conduct a research interview?
Watson may get there, but I remain convinced that the biggest challenge for Watson may be learning to ask the series of contextualizing and narrowing  questions that must follow a simple and common question such as “What is the statute of limitations for breach of contract?”  Can a lawyer accept a google-ized /wikipeidi-ized generic answer which is not curated  to address the specific facts of his client's situation?  We all know the frustration of hearing this standard Siri refrain “I don’t know 'mens reä.' ”  Legal advice requires a higher level of precision than a Jeopardy-style fact based query. This is not a slam dunk.

A Humble Suggestion  I just suggest that Watson’s developers  find some research librarians who have served serious “hard time” at a large law firm reference desk to run Watson through the paces. If Watson is to master legal research it needs to learn how to ask questions from the pros!

No comments:

Post a Comment