WatsonLaw Next?

Thomson Reuters and IBM announced earlier this month that they will be joining forces to “enhance customer solutions across Thomson Reuters using Watson.” A very interesting, if not a somewhat inevitable, development in 21st century legal research.

In the press release Mike Rhodin, senior Vice President of the IBM Watson Group enthusiastically said:

“Working with Thomson Reuters, and their vast trove of data, is an incredible opportunity to combine Watson’s cognitive capabilities with a global leader in decision making solutions across science, legal, tax, and finance. The result will be accelerated discoveries for the professionals that rely on these important information solutions, ultimately bringing new levels of speed and precision to critical decisions.”

I had the opportunity to attend a lunch session on IBM Watson that Gale Cengage hosted in January. It was a great introduction. In my notes I wrote: “don’t really know what’s coming next, but need to seize on this opportunity, you plus Watson, the assistant who knows all of the research/knowledge in your domain; humans can’t do this alone, ability, biases, etc.” Although, it seems I also commented: “fascinating and terrifying.”

So what do you think? Are you ready for your new legal research assistant?

And, for obvious reasons, and certainly with no intended disrespect for this initiative, I cannot help conjuring up the HAL 9000 whenever I think of IBM’s Watson. So I am compelled to leave you with one of HAL’s many pithy yet soothing quotes:

HAL: I know I’ve made some very poor decisions recently, but I can give you my complete assurance that my work will be back to normal. I’ve still got the greatest enthusiasm and confidence in the mission. And I want to help you. —2001: A Space Odyssey[1]

 


 

[1] Courtesy of IMDb http://www.imdb.com/character/ch0002900/quotes.

Comments

  1. As exciting as these developments are, lets not lose sight of the fact that Watson or any other cognitive technology is only as good as the content or information that goes into it. Garbage in, garbage out — this adage applies to the human mind as well. Technology may have cognitive ability but for those about to despair there is still hope the human gifted with the abilities of independent thought and analysis may be even more in demand.

  2. I see a bright future, not far from today, where Canadian Law Societies band together to invest in Watson-based solutions to prepare legal briefs directly for clients, who would then take those briefs to barristers to review them and present them in court. And if the Law Societies won’t buck up, private industry should provide the service, again, directly to clients (as legal information, mind you, not advice). Plug Watson into the CED and CanLII and make it go a few rounds with top counsel and I think the result will be better than most expect.

    This is keeping in mind, however, that lawyer oversight will always be required. Remember Watson’s Final Jeopardy blunder. Question: “Its largest airport was named for a World War II hero; its second largest, for a World War II battle.” Watson’s answer: “What is Toronto???”

  3. At the Pacific Legal Technology Conference at the start of October, we debated how fast Watson would make an impact on the legal market. I pointed out how messy much legal information is, and that large amounts of secondary legal information – as well as deep corpuses of primary information – were controlled by Thomson Reuters, Reed Elsevier and Wolters Kluwer (for the Canadian market, I discount Bloomberg at this time).

    But in the debate, we also said that if either Westlaw or Lexis could harness Watson to feed the next generation of search, then the combination could be revolutionary.

    It won’t be a fast challenge, but this announcement is a potential game changer, especially if there are exclusivity clauses in the collaboration agreement.

    Expect the midnight lights to be burning in Eagan for the foreseeable future.

  4. This would have been excellent rebuttal material. (As the other voice in that debate on Oct 2 against Simon, I should know!)
    Thomson Reuters’ corpus of knowledge is kryptonite for the argument that Watson won’t progress because no one would ever want to seed it enough law data. As for “garbage in, garbage out” — and obviously Thomson Reuters is not contributing garbage—I would say that even if bad info was inputted it is only true for a time that Watson would spit out what it was given. Watson learns by questions sets and answers sets. It looks for patterns therein. Thomson Reuters has spent several years amassing massive stores of not just caselaw and commentary, but the precursors to caselaw—namely counsel’s argument contained in memoranda of argument on both side of an issue. Armed with an abundance of precedent caselaw and commentary as anchor points, Watson would set to the task of finding patterns between: a) counsel’s arguments; and b) results delivered by judges.
    Watson would be able to do this with the 1,000s of files Thomson Reuters copied into its Litigator database. Once strong patterns emerge, the system’s ability to filter “garbage” inputs would improve.
    We’re looking at AI’s possible “impact on the legal market”, so we should address our honest tendency to think along the lines of “how will AI change the ways I as a lawyer serve my paying legal clients “. We put ourselves at the centre of this mental exercise in forecasting what impact AI will have. I don’t think it’s the best starting place, nor does it produce the best reasoning. We should be asking how will AI allow clients to sidestep the traditional legal service sector by using a cheap software alternative (owned by non-lawyers). And if we are really interested in impact we should pull scope to consider not just the existing market for legal services, but markets that the profession struggles to serve at all—legal guidance or dispute resolution for small businesses and even the middle class.
    When it comes to defining “legal services”, we get caught up in equivocation. AI researchers have complained that whenever AI improves, humans just move the goal posts and slice away at the definition of what “intelligence” actually is. We used to say that when a computer can beat our best human chess master, AI will have arrived. That happened in 1997, and we quickly dismissed that feat of narrow intelligence. In legal, we will keep redefining the quintessence of what lawyering means. E.g. as auto-correct, voice recognition and proof reading software can now parse our inept verbal spewing into coherent sentences, we still would not call that an impact on lawyering—just some clever programming and algorithms. And what about predictive coding in e- discovery? Or filling out court forms? Oh, those are things a summer student could do, so not convinced.
    Any time AI masters a process, we step back and redraw the line between neat trick and skill. Will this be the case when we ask Watson for a memo and get a list of likely legal propositions with confidence scores and supporting authorities? Will we just call that augmented data filtering? When Watson processes an intake survey for a client and churns out a letter of demand, pleadings and a list of questions for discovery? Will we just call that low end junior work?
    Where is the real line on abstract/analogical reasoning and problem solving/negotiation/persuasion skills? What is the “je ne sais quoi” lawyerly cognition quintessence? If we are not prepared to see it on the horizon, are we even prepared to see it ever?
    We should examine our confirmation bias. Lawyers are not immune to arguing from consequences. This looks something like: “If we agree that thinking machines may soon be capable of doing our work, lawyers may be replaced. That would be terrible for lawyer morale, therefore AI will not soon become capable of doing a lawyer’s work.” If we only seek evidence that confirms our per-held views, this does not solve the pure question of whether or not machines are actually on a fast track to thinking like lawyers.
    While lawyers have a monopoly on legal services, is there really the guarantee that IBM working in conjunction with Thomson Reuters could not produce some kind of legal AI software assistant that could provide legal opinions directly to clients? Could ABC Ltd. not licence an “in-house counsel” app to provide the company advice about employees, secured lending, contract review and negotiation tips, privacy law compliance, etc.?
    Private investment in the AI sector has been expanding 62 percent a year since 2010. Not all (or even much) of the $17 billion invested in AI since 2009 has been in legal applications, but we must appreciate that as AI’s capacity grows and disrupts bigger markets, the day will come when the legal knowledge industry will be no match for an evolved AI.
    Looking at BC’s Legal Profession Act, I see that s. 15 prohibits a “person” from practicing law. It says nothing of machines… but should it? Is it now important for Law Societies to consider how we might preserve the public interest in the event some company opens up an online legal portal for the public using Watson?
    There has been a lot of debate over what adverse effects alternative business structures and ownership models will have on the core values of the legal profession, including lawyer independence, client confidentiality, and the duties owed by lawyers to a client – particularly the duty to avoid conflicts of interest. But is it time to consider the impact that AI will have on all of these things?
    For conflicts, consider that Watson is essentially a hive mind of information… thousands of servers yoked together. Is there any walled garden in place to prevent data from one file being used in another? Can Watson cone of silence itself? Could two firms, both augmented by Watson and seeking strategic advantage, work on opposite ends of the same file?
    Once we settle the question as to whether AI will impact the legal market, there are so many other interesting questions.

  5. Thanks everyone for your great comments!

    Nate, I just wanted to pick up on your last points on confidentiality and public access.

    The question of confidentiality and Watson was raised during the Q&A of a presentation given at the American Association of Law Libraries conference in July.* Kyla Moran, from the IBM Watson Group stated that there was no “cross-pollination” between Watsons. Each instance is fed, trained and operates independently. Watson does not come pre-loaded with the latest legal information although I could see this as a selling point for “WatsonLaw Next.”

    As far as making this publically available that might be something that emerged in the future. It seems they are, at the moment anyway, primarily interested in “linking smart people with smart machines.” I take that to mean providing intelligent assistance to professionals like doctors, lawyers, scientists, etc. In this sense she stressed that Watson is an “augmented intelligence.”

    Having said that, check out the public instance of Chef Watson for some “cognitive cooking.”

    * If you’re an AALL member you can watch the presentation here.

  6. “Armed with an abundance of precedent caselaw and commentary as anchor points, Watson would set to the task of finding patterns between: a) counsel’s arguments; and b) results delivered by judges.” Much is based on precedence; however, when recognizing patterns is there a suggestion that every case scenario and its decision has been predetermined? I realize that there is already much reliance on predictive analytics and programs. Is it as scripture states: “What has been is what will be, and what has been done is what will be done; and there is nothing new under the sun” (Ecclesiastes 1:9)? If such is the case, then perhaps lawyers and courts are already redundant.

  7. I would not be the one suggesting that patterns in caselaw promise determined outcomes. But cognitive computers should be capable of assimilating these patterns and reporting on probabilities. David Canton wrote a comment on Slaw earlier this year: “Watson and computer intelligence in general is going to whack lawyers on the side of our heads. Too many lawyers wrongly insist that their knowledge/advice can’t be reduced to an algorithm.”