I seem to have fallen off the blogwagon lately and am now attempting to turn my mind back to some writing. So I’ll start by reporting on one of the sessions I attended at the recent Canadian Association of Law Libraries conference held in Vancouver, from May 15 to 18. The session took place on the afternoon of May 16 and featured: Steve Matthews, Slaw publisher and contributor and founder of Stem Legal Web Enterprises; Ivan Makonov, Executive Director at Lexum; Eric Laughlin, Managing Director of the Corporate Segment, Thomson Reuters; and Nate Russell, liaison lawyer with Courthouse Libraries B.C., Slaw contributor, and manages Clicklaw. The session was called “Computers in Legal Research” and the panelists set out to explore the role of artificial intelligence (AI) in legal practice (“is it really possible for a machine to replace a lawyer?”) and the implications that these changes might have for law librarians.
The panelists started things off by framing the current state of AI. There has been significant development in the field since IBM’s Deep Blue, in May 1997, became the “first computer system to defeat a reigning world champion in a match under standard chess tournament time controls.” Fast-forward to 2011 when IBM’s Watson, the “question answering computer system” that beat Jeopardy champions Brad Rutter and Ken Jennings.
And more recently, just a few months ago in fact, Google’s AlphaGo became the first computer system to beat 9-dan professional Go player Lee Sidol. One of the interesting things about AlphaGo is that it was trained first by a human and then learned further strategies by playing games of Go against itself.
In the legal arena Neota Logic have been working on legal reasoning applications to provide their expert question/answer systems that includes AI solutions for legal aid. Another expert system developed at the University of Toronto is ROSS. ROSS Intelligence uses the IBM Watson cognitive computing technology and has recently partnered with Baker & Hostetler, a large global law firm specializing in litigation, business, employment, IP, and tax. It will be very interesting to see what develops out of this partnership.
Makonov offered some insights into the evolution of AI boiling things down into these three definitions:
- narrow intelligence – used for specific things or tasks, e.g. stock trading algorithms
- general AI – a human like system, capable of problem solving, reasoning, planning and learning like a human
- super intelligence – a system greater than the sum total of all human brains on the planet
Yeah, so that last category led naturally into a discussion of AI and the fear factor.
But first Matthews talked a bit about the specifics of what AI can do. This includes removing some of the mundane, routine tasks that are part of most of our professional lives. The upside of that is this potentially frees us up to perform the things that us non-machines do best: analyze, provide context, draw from our real life experiences and apply them to the situation at hand. In that sense the ‘A’ in AI is better thought of as standing for “augmented.” That is, AI augments our intelligence and serves as a “tool, not a competitor.”
Makonov then touched on the “developer’s perspective.” “Software,” he said, “can be superior to humans in many ways … [but] AI is just another thing that software can do.” As technology users we are exposed to the “application layer.” Developers, on the other hand, build on software “platforms” to applications that deliver specific functions. He equated the non-developer’s perception of the mysteriousness of how software works to the once equally mysterious understanding of electricity when Thomas Edison introduced electric light to the mainstream in the late 1870’s.
I believe it was Laughlin who reviewed some of the things that software can be particularly good at, for example, document review and client intake. He also mentioned Blue J Legal, a system that analyses fact situations returning relevant case law and other materials and succeeds in reducing the mysterious aspects of the process by revealing how the system arrived at its results. This is AI that delivers answers, it’s not just an information storage and retrieval system that helps find documents in your database. It’s a system that can evaluate legal strategies, using evidenced-based processes, and is able to “play the chess of law.”
Librarians may have to up their game a bit when working with this emerging field. What can librarians do to upgrade their skills? Matthews suggested re-engaging as collection builders, cataloguers, and knowledge managers. Now, more than ever, the need to understand the organizational cultures that we work in is very important. This understanding will help better develop and communicate a vision of how information is used in your institution. We should realize that we are product experts who know how to use information and remind ourselves that in the end someone will need to manage these AI resources.
We should continue to improve our tech literacy, by becoming familiar with software coding principles, data management, text markup, etc. And we should be prepared to give up some of our traditional turf, for example some of those many mundane tasks that we may no longer have to shepherd as we go forward. We should position ourselves as advocates and not fight against these changes. Fortunately these are all the types of things that librarians as a profession are naturally predisposed to do.
However, Makonov left us with this to think about: “If you’re not at the table, then you’re on the menu.”