Artificial Intelligence in Law: What Are the Consequences for Law Librarians?

Several Slaw contributors have written recently about the use of artificial intelligence in law (Tim Knight here, Nate Russell here) with particular reference to the program on “Computers in Legal Research” at the conference of the Canadian Association of Law Libraries held in Vancouver this past May, moderated by Slaw’s Steve Matthews. I attended the program. I was disappointed though not surprised that none of the speakers was a librarian; and, while there was much discussion of the potential and possible consequences of artificial intelligence (AI) in legal practice, there was, aside from the moderator’s comments, no discussion of how AI might affect law libraries. So I’ve been asking myself: what are the possible consequences of AI for law librarians?

AI, Watson, ROSS and Legal Research

Every reader of Slaw will have heard of ROSS, being developed here in Toronto by a group of former University of Toronto students. Built on top of Watson, IBM’s cognitive computer, ROSS is a digital legal expert designed to power through legal research. ROSS claims its language processing capabilities allow it to respond “intelligently” to questions posed by lawyers. You ask your questions in plain English and ROSS then reads through the entire body of recorded law, gathering evidence and drawing inferences about the materials it has evaluated. It then returns a relevant answer with cited references and topical readings from legislation, case law and secondary sources to get you up-to-speed quickly. Further, once the research is done, ROSS will keep you up to date on new court decisions that could impact open matters. Similar to Watson, ROSS learns from its interactions and reportedly provides better results after each use.

The law firm Baker & Hostetler has recently introduced ROSS as the “new associate” in its 50-lawyer insolvency practice. The press release from the firm insists only that “emerging technologies like cognitive computing and other forms of machine learning can help enhance the services we deliver to our clients.” There is no indication of what those enhanced services will be.

In view of ROSS, it is only natural that Thomson Reuters, the owner of Westlaw, is collaborating with IBM to integrate cognitive computing (ie, Watson) into its “customer solutions” and has established a data and innovation lab (Thomson Reuters Labs) in Waterloo, Ontario (Canada’s Silicon Valley) to “bring together our unrivalled information assets and industry expertise with experts who can help make those assets increasingly actionable for our customers.” It is certain that, in time, ROSS will not be the only AI-enhanced legal research service on the market and that AI will move into enterprise-level, in-house applications for knowledge management (KM) and practice management.

As a rule, technology has been used in industry not to enhance service but to realize economies. By implementing ROSS and sparing its lawyers the onus of research, a law firm should be able to handle more matters in the same amount of time, increasing the lawyers’ productivity and the firm’s profitability. Alternatively, with the competitive advantage provided by ROSS, a firm could lower its fees without reducing profits, a definite service to clients. Another possibility is that, with improved productivity, the firm would be able to maintain levels of profitably with fewer lawyers.

AI, Law Libraries and Law Librarians

AI in legal practice is not a “Future Shock” scenario. It is real, here and now. Richard Susskind has written several books on the implications for lawyers and identified several developing career alternatives – legal hybrids (lawyers taking on an added discipline as strategists, management consultants, psychologists, etc), legal project managers, legal risk managers and others. But what about law librarians? What are the chances of law librarians being made redundant by Watson or ROSS, and what are the qualities and skills that might allow librarians to work with AI rather than be replaced by it?

If ROSS or another AI-enhanced application can conduct legal research intelligently, eliminating the need for lawyers to do it themselves, how much legal research training will lawyers need? Most lawyers would be perfectly happy to let ROSS do legal research for them. If legal research is no longer a core competency for lawyers and provides them with no competitive advantage, should we be reconsidering now the need for dual-competency lawyer-librarians in our law school libraries whose primary responsibility is legal research instruction? At a panel held at Queen’s University law school in Kingston, Ontario, last year, on “How will artificial intelligence affect the legal profession in the next decade?”, Jordan Furlong (another Slaw regular) made the following observation:

“I’m inclined to think that a lousy lawyer with instantaneous access to all the legal information in the world is still a lousy lawyer. You still need to have the qualities we consider to be outstanding features of good lawyers, which include analysis, empathy, good judgement and all sorts of different things that Watson doesn’t pretend to have. But those are the qualities that are going to mark the lawyers who can work with Watson rather than the lawyers who will be shoved out of the way by Watson.”

Legal research skills may not be one of the qualities of lawyers who can “work with Watson”; while legal analysis, legal reasoning and client skills will be essential. Should the emphasis of LRW (Legal Research & Writing) programs in law schools be shifted to “Legal Reasoning & Writing” instead? And if there is a change if focus, will we still need librarians to teach in these programs?

Will law firms still need librarians for legal, business, competitive intelligence and current awareness research? If ROSS is good and keeps getting better, could it be that librarians’ research skills will no longer be enough to keep them as partners in or supports to legal practice? At that same panel discussion at Queen’s, Jordan Furling also made this comment:

“Data has been an issue for the law for quite some time. If you’ve worked in a law firm, you know how incredibly difficult it is to acquire, collaborate on and pull together knowledge in a meaningful and applicable sense. For a while yet the technology itself will probably be some distance ahead of the legal community’s willingness or ability to provide it with the information it needs, the fuel it needs to run.”

Artificial intelligence systems like Watson and ROSS, no matter how sophisticated their computing algorithms, will ultimately depend on good information to be successful. Information is what librarians “do”. As “information scientists” (to use a dated term), librarians are trained in the skills and also the arts of retrieving, analyzing, describing, organizing and structuring data. Added to these hard skills are the soft skills that have always been signatures of the library profession: teamwork, collaboration and a commitment to service. Librarians also realize the importance of context, ie, the need to understand the organizational cultures in which they work, adding to a better understanding of how information is used in their institutions. These skills foster a professional culture that attunes librarians to “acquire, collaborate on and pull together knowledge in a meaningful and applicable sense” (as Furlong described the process).

Law is a profession driven by information. As Steve Matthews commented in the program mentioned at the start of this column (as quoted by Tim Knight in a previous post), if librarians are to remain relevant in an environment where access to legal information is mediated by AI, they must re-engage with traditional library skills as collection builders, cataloguers, indexers and knowledge managers. But it will not be a cake-walk. These traditional library skills are not generally available or even promoted in academic law libraries, at least in Canada, where the “technical” aspects of law librarianship have generally been outsourced to central libraries and the focus is almost exclusively on legal research and reference. All librarians will need to improve their technological literacy, especially coding and text markup, better to understand and assist with the transformation of information into data. In both law schools and in law firms, librarians will also be in competition with lawyers seeking alternative careers in law as “legal knowledge engineers” (as Susskind calls them).

Is there a role for librarians in law post-AI? There can be if we begin now to prepare for that future by acquiring the skills that future will require, by focusing not on research but on information structure and systems, and by working with and not against the inevitable changes.


  1. As a Librarian moderator for the CALLACBD session, I tried to interject some comments along the way. What this comes down to for me is, how quickly do we see sentient learning of the firm’s resources taking place? Personally, I can’t conceive of this being less than a 5-10 year proposition. And even if AI produces this type of sentient learning, developing original bodies of organizational knowledge will continue to be a difficult task that requires some element of creativity.

    That leaves an opening for Librarians to become the “experts” with how AI software systems engage with *their* firm’s resources. I say “their” because AI should become not just a ‘value add’, but a competitive advantage that helps differentiate legal services. I see a future environment where Firms will buy or build different in-house AI engines, and clients will measure their legal counsel by how easy that engine is to deal with.

    How do you make your firm’s AI engine better? Easy: Load it up with quality resources, and provide a better answer than the firm across the road can.

    That requires a deep dive into the legal needs of the businesses you’re serving (lawyers), and it requires aggregation of a firm’s knowledge and experience (librarians, KM, engineers, coders….).

    Why should firms be investing in librarians and KM? Because if they don’t, in 10 years time, their AI engine will suffer from “GIGO” (Garbage In, Garbage Out). Having limited (or no) inputs that reflect your institutional expertise, in any service business, is going to seem short sighted.

  2. Kind of a sidebar, but how is “information scientist” an outdated term? In a world where euphemistic monikers are proliferating (lawyers as “legal knowledge engineers”, librarians as “knowledge management professionals” etc.), are law librarians wanting to keep their designation pure?
    A 2014 report called “At a Tipping Point” grappled with the challenges of the brand “library” (and by extension librarians) and the risk that the brand is losing relevance. Libraries = Books, is literally what the report’s research about public perception says. Even as libraries dramatically increased spending on digital resources over the last 10 years, the mindshare has not shifted.
    If the future for librarians is to enhance technology skill sets, even coding, in order to operate or merge (in a manner of speaking) with AIs (the term “centaur” is a thing in chess where humans use AI recommendations for play to beat pure AI or pure human opponents), then maybe “information scientist” is looking more relevant.
    In any event, although I’m not a law librarian, I have been embedded in a law library for over 5 years and find this to be an interesting — and critical — existential debate.