I, Reference Robot

Last week, the computer named Watson took on two of Jeopardy’s all time champions in a two day match. Developed by IBM, the computer was designed with the intent of attempting to respond the unique “answer first” trivia format of the show. It trounced the human opponents on both nights of the challenge and not by small margins either. And with its win, it raised the possibilities of what the computer could do in the next generation.

I was rather surprised at the muted reaction of the librarian blogosphere. With the exception of a post at Henderson Valley Eggs, there wasn’t any sort of commentary. Given the glimpse of capability that the computer like Watson represents, I would have hoped that there would be a bit more excitement about the possible library applications. While Watson is probably not ready for prime time at the reference desk (due to how the program was deconstructing the clues), I could not help but marvel at the potential.

My excitement in getting a computer like Watson in the library is having a tool that can handle known requests (as in “I want Cross Fire by James Patterson” or “What books on biology do you have?”), some harder requests ("I want the movie that came out last year with Stallone in it” or “Can you tell me what order the Lillian Jackson Braun books came out in?”), or just requests that require deep scanning (“It’s a book with a red bicycle on the cover and it’s about an aunt’s suicide”). I’m not sure that the computer would be able to handle all types of requests, but I think on a long enough timeline it would be able to handle complex speech. (I’m trying to imagine if it had Google’s voice recognition data that it gathered from its Goog-4-1-1 information access number; now that would be pretty awesome.) But, in the meantime, I don’t believe there would be shortage of questions it could answer. Think IM, text, and chat questions; it runs a search, shows it to a human operator who approves or corrects, then the reply is sent. The turnaround for easy questions could be under a minute; corrections or harder answers could be not much more.

As I called it a tool, I don’t see it as a replacement for any library staff. As such, I can see it freeing up staff so as they can be able to offer more programs, services, or be able to have time for additional librarian work both at and away from the desk. It’s a great technology and I would hope that one of the future challenges that would Watson’s programmers would take up would be trying to handle reference style questions. For those who may balk at a computer with this kind of capability, I personally feel that it is an inevitable technology; as such, rather than shunning it, it should be embraced and integrated as soon as possible.

I’ve linked the library scene from The Time Machine below as an idea of future capability (even if the hologram needs some serious customer service training).

21 thoughts on “I, Reference Robot

  1. I think people would be intimidated by it, and probably want to talk to a real person. But somewhere down the road, sure, why not?

  2. Andy,

    I love your idea as a tool. As a Children’s reference librarian, I get a lot of “it had a bus on the cover.” Fortunately, with Amazon, Novelist, and almost 15 years in the book business (8 in libraries), I can usually come up with an answer. I would love to see a professional article on how Amazon is used by so many librarians because you can severely misspell a title and still get to where you wanted to go. (So crucial with sci/fi titles.)

    While I don’t have my finger on the pulse of library blogs as you do, I’m also surprised that there wasn’t more library buzz on Watson. I don’t watch Jeopardy and my parents were traveling this week, so I only caught the buzz through a few tweets, the Slate article by Ken Jennings and the great story on CBS Sunday morning.

    And now I think I need to watch that movie!!

    Thanks for asking the questions to get us thinking, Andy. Keep up the good work. And here, have a cupcake.


  3. I was really excited about Watson simply because I love Jeopardy (I got to watch College Jeopardy at UW-Madison during grad school). It was also an exciting milestone in AI and programmable intelligence. Watson could learn from examples of older Jeopardy questions and partially decode human language and wordplay.

    I don’t see it as a reference robot. First, when people come to the desk and ask a question that is often not really the question they are asking. When they say “I’m looking for information on the Hippocratic Oath,” they may be looking for nothing of the sort. They may be trying to understand ethics in medicine but can only think of the Hippocratic Oath. Humans can dig deeper and get to the question behind the question.

    Also the experience that people have in libraries is important and a human can deliver much better customer service than a robot. They can empathize with you when you get frustrated. “Yeah, that’s lame. I can’t believe we don’t own it. Let me get it through ILL.” Though personally I would love talking to a robot (because robots are awesome) a human is going to give a much better experience.

    But perhaps this technology can be used for categorization or cataloging or decoding natural language when trying to organize information. I’m not sure exactly how that would look, but I could see that as a distinct possibility.

    • Andy, thanks for your comment. I was careful to nuance my post about what I thought a computer like Watson can handle; it certainly would not (at current ability) be able to handle some of the inquiries that you mentioned. Nor the fun ‘I’m looking for this but I’m really looking for that’ reference discovery.

      Yes, there are going to be times when the human will do better than the computer. But I see it the same way that there are self checkout stations; an additional avenue of inquiry that is available for those who want it. I wouldn’t hazard a guess as to how quickly something like a reference robot might develop; they built Watson in four or five years, from what I read. It’s certainly a wait and see, but I’m an optimist.

  4. Pingback: Tweets that mention I, Reference Robot « Agnostic, Maybe -- Topsy.com

  5. I was quite excited by Watson too. I have no doubt 5 years down the road our vendors will be trying to sell us Watson like technology has the next big thing… (New gen catalogues, Federated search, discovery services(

  6. Pingback: Should we welcome our reference overlords? | Robosaur

  7. Hi Andy,

    I was in the process of writing you a comment here, and once I passed 500 words, realized I should probably turn it into a blog post of my own instead. ^_^ In brief, I share your excitement for the technology, but thing that it will be significantly disruptive for the library profession, and that we would be well served to consider Watson’s victory a “Sputnik moment”.


  8. I watched the first 2 games. What impressed me was that Watson blew the final Jeopardy question about U.S. cities. The answer was something to the effect about a U.S. city with 2 airports: one named for a WWII hero and one for a WWII battle. Watson answered “What is Toronto?????” ( the multiple question marks denoted an unsure answer). The correct answer is Chicago which I deduced by knowledge that Chicago has an airport named Midway which was a WWII battle.

    I wondered how a computer programmed with so much information could make such an elementary mistake about the location of Toronto which is in Canada the last time I looked. Until a computer can make the kind of connection that I made, I don’t think it’s ready for the reference desk.

    • I wouldn’t put that as a disqualifier from service since librarians (or people in general) can make mistakes like that, but I would be interested in hearing how it came up with that answer.

  9. Great post. What caught my eye was your wish about “getting a computer like Watson in the library” and the robot image, which implies a physical presence.

    I agree that something like Watson would be a great tool for answering reference questions or providing ideas for read-alikes etc., but I assume this would be something more like an online service, where ‘Watson’ exists on a server somewhere and is actually processing lots of information requests from libraries, and other points, all around the world.

    Designing an interface that folks feel comfortable with is probably the tricky part. A hologram or even a CGI talking head has got to be better than talking to a box.

    • I think there is also a novelty factor to it, but I’m sure that would wear off quickly if it was incapable of answer a person’s question.

      As we move towards user experience, yes, the interface will be pretty key.

  10. Have you seen the 1957 movie “Desk Set”? It is quite relevant to Watson’s library usage. I think that in this period of cuts and layoffs the entry of Watson and Watson-like computers scares librarians as much as the appearance of the computer in the movie. Not saying good or bad, but just as unsettling. Librarians will need to continue to develop new skills and knowledge sets if Watson comes to work with us.

  11. The pros and cons of being able to eliminate positions aside, let’s not kid ourselves. The time saved when automation is introduced never ends up being saved by individual staff members who are freed up to do higher quality work. That time saved (whether the transition happens the next day or slowly over a year or two) ends up being time saved from the payroll.

  12. I think we’re getting a little ahead of ourselves, here. In this discussion of Watson as a “digital associate” that might assist (but not replace!) the reference librarian, we’re ignoring logistics. This computer is enormous. This computer is expensive. Libraries can’t afford to stay open on Sundays, and we don’t have enough space for our books. This computer is not coming to a library near you anytime soon. Maybe in fifty years, but certainly not in five.

    Who does have the space and the money? Google. It might have been more apt for Ken to welcome our new commercial overlords.

    My guess is that the technology driving Watson is going to revolutionize search in a way that might seriously affect the relevance of reference service in libraries. I don’t mean to be pessimistic, and I don’t think this is the weapon Google needed to do us in once and for all, but I just don’t see this as a boon for the reference desk.

  13. Pingback: “Hot Topic” of the Week – IBM and Watson « Against-the-Grain.com

Share your thoughts

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s