Nicholas GallMy colleague Mark Raskino has written a thought provoking piece (Could recession speed up lawyer automation?) in which he predicts "‘expert systems’ are coming into legal practice situations. These systems can take away much of the simpler work previously undertaken by legally qualified people at significant hourly rates."
This mention of "expert systems" replacing attorneys brought back fond memories of my youth. One of the events that led me to switch from being a software developer in the field of AI to becoming an attorney (I entered NYU Law School in the Fall of 1988) was my attending the First International Conference on AI and Law (ICAIL) way back in May 1987 — over twenty years ago. I figured that if the attorney thing did not work out, I could always take what I learned in law school and use it to write a legal expert system to replace attorneys. Ah the naivety of youth…
It’s interesting to note that one of the presentations in 1987 was by Richard Susskind, who Mark credits with the idea that expert systems are coming into the market. The name of Susskind’s 1987 presentation? It’s "Expert Systems in Law: Out of the Research Laboratory and into the Marketplace." Well, I guess it’s been a very slow entry into the market. Here we are almost 22 years later and I don’t think expert systems have even begun to make a dent.
And I’m not sure they ever will. Or let me put it another way, by the time expert systems make rank and file attorneys worried about their livelihood to a significant degree, most other white collar workers will have long been automated out of existence — including IT Industry Analysts. Most white collar work is far more susceptible to automation than legal work. Legal concepts are some of the most abstract and logically difficult to formalize.
How in the world is an expert system going to formalize concepts like mens rea, consideration, or fair use? To this day Westlaw uses masses of bar-certified attorneys to read every court decision and to semantically annotate the text paragraph by paragraph with Key Numbers. Westlaw has done extensive R&D to find ways to automate such annotation, but after decades of trying, the best they can do is to give the human annotators some automated assistance.
It’s interesting to note that the ICAIL conference is still going. Wow, those AI researchers never give up! But the latest conference, ICAIL 2007, has nary a mention of expert systems, which is troubling if they are poised to disrupt the legal profession any time soon. Instead, virtually all the presentations seem to be about legal ontologies. No the white collar workers in the legal field who have the most to fear from automation are paralegals — they’re the ones already doing all the routinized legal tasks, like bates numbering. Shudder…
The one thing about the practice of law that might be a casualty of this recession is hourly billing. When the presiding partner of Cravath, Swaine & Moore, Evan Chesler (for whom I had the privilege of working as a summer associate oh so many years ago) says its "time to get rid of the billable hour", it seems almost possible that this sacred tradition might come to an end. At the very least, billable hours will end long before expert systems decimate the ranks of attorneys.
BTW, when I finally left the practice of law, I did take a crack at writing some legal software. I tried to develop a program that would automatically generate legal cites in Bluebook form. Seemed simple when I started the project, but once I got into it I realized it would take almost full blown AI to do it completely and reliably — so I gave up and became an analyst instead.
1 response so far ↓
1 Galen Tosham September 30, 2009 at 10:08 am
“Value billing” whether it looks like the billable hour or not has been the guide to many who actually have to write the bills and send them to clients.
“Law”, of course, it’s not a monolithic entity — there are many widely differing sorts of tasks involved in the numerous kinds of practices the term encompasses. But the order of change seem right to me here — it would be interesting to see what sort of cross examination would be the result of automation say, of a defendant in a civil case who is beginning to look grandfatherly and enthusiastic in trying to get the jury to understand some technical issue when the theme of the case by the plaintiff appears to include that he is an egomaniac though it’s just after lunch and some of the jurors look sleepy and in need of a bit of waking to focus their attention, but the Judge appears to be in some sort of a mood and it might jeopardize a few critical upcoming motions in limine to prod the witness into an angry outburst and jurors 3 and 5 looked a bit uncomfortable the last time anyone raised their voice questioning a witness and you’ve got another half day of cross and it might be good to save drama for the end when the jury might better remember it…or perhaps it’s more memorable when it fits in the timeline of the narrative …