August 20, 2017

Promise and Peril of AI for Academic Librarians | From the Bell Tower

Steven BellIt is still early in the development of artificial intelligence but eventually it will change the work of librarians—or make it irrelevant. How likely is it that we will be replaced by bots in the future?

Artificial intelligence (AI) took a big leap forward on March 9, 2016. That’s the day the world’s master of Go, a complex board game, was badly beaten by Google’s DeepMind Alpha Go computer. Lee Se-dol exclaimed in surprise, because he never thought a computer could beat him at Go. Computers beat humans at chess and Jeopardy long ago. Go was seen as a vastly more rigorous test of artificial intelligence because it requires instinct, intuition, and evaluation capabilities, human qualities that computers cannot match—or so we thought until now. Leading up to this match and in its aftermath there has been extensive speculation in the mass media about a future where artificially intelligent robots and devices interact with humans. Will they free us to enjoy life more fully? Produce Matrix-style enslavement of our race? Simply take over most of our jobs? While the impact of AI is hard to predict, experts agree it will happen faster than we expect.

Are Librarians Replaceable?

Industrial robots have already eliminated thousands of workers’ positions by performing their manufacturing jobs. It is now a question of whether robots can do the jobs once thought safe from being easily replaced. In his book A Whole New Mind Dan Pink’s advice to avoid being made obsolete by outsourcing, offshoring, and what we’ll probably refer to as “getting roboted,” is to gain employment in professions where “right brain qualities—inventiveness, empathy, meaning—predominate.” In other words, soft skills that require an emotional intelligence are exceedingly difficult to automate or delegate. When I first read A Whole New Mind I was reasonably sure that much of what library workers do, particularly when it involves design activity, fits well with those “right brain” activities that should make us irreplaceably robot-proof. Now I am less certain.

AI Mania

Even before the battle of Go I had noticed a distinct uptick in online news and debate about a future shaped by artificial intelligence, and whether it would lead to a new era of freedom for humanity or enslavement by robot overlords, with frequent references to the Terminator and Matrix movies. Articles like this one question whether our jobs will be eliminated by artificially intelligent robots. We should be safe for the next ten to 15 years. It may take that long for image recognition and data analysis to advance sufficiently for application to low-level human tasks. But don’t let that 2015 DARPA robot challenge make you overconfident about your job security. AI will change the job market. In the not-too-distant future deep learning technology will allow a machine to recognize patterns and use trial and error to teach itself a how to accomplish a new task—kind of the way we do. Geoff Colvin, author of the book Humans Are Underrated, believes we have no reason to worry. AI’s mastery of complex games only suggests, according to Colvin, that “the skills of deep human interaction, the abilities to manage the exchanges that occur only between people, will only become more valuable.” Skills like empathy or creativity should allow humans to maintain a critical advantage over AI, in that they allow us to better understand the irrationality and unpredictability of our fellow humans.

That Robot Librarian

Do all library jobs need those skills? Could artificially intelligent machines eliminate library jobs? Consider a self-checkout terminal. It supplements the work of humans, although it lacks intelligence. But it’s not inconceivable that a robot could perform any number of similar library transactions and have the intelligence to respond to a human query. That seems to be the idea behind Hugh, the the robot librarian created by some researchers and engineers at Aberystwyth University. Librarian is probably not quite the right term since all Hugh can do is search the catalog, identify a book’s shelf location, and lead a patron to it. Still, if it can do all that it’d be impressive. Though not much more than a novelty for now, a robot library clerk could potentially replace stack or circulation workers. Hugh might be just the beginning. That’s the promise of AI for librarians. It could allow us to dispense with conducting routine transactions so that our resources could fully support complex interactions.

Where Does It End?

Based on what we know today about AI, despite all the excitement over a game of Go, academic librarians’ jobs should be secure for the foreseeable future. I wonder if the next generation or two of academic librarians will have that level of security. If today’s AI can be programmed for intuition and evaluation, it’s hardly a stretch to imagine AI being applied to question interpretation and resource recommendation. The things AI systems could learn about us as we conduct research could help humans to achieve better results. It’s conceivable a research project could be launched with a just a simple request. Today you can ask your phone’s digital assistant for directions or movie times. Tomorrow, why not a complex research query? Faculty, researchers and students all depend on librarians for their knowledge of sources and ability to navigate complex research resources. Minus the empathy, AI could potentially lead humans to make the discovery. I can see why some people are getting excited about a new age in which AI could free humans from the burden of tedious work—and be our robot servants. Librarians will need to learn much more about AI to understand where the line will be drawn between servant and replacement.

Steven Bell About Steven Bell

Steven Bell, Associate University Librarian, Temple University, Philadelphia, PA, is the current vice president/president-elect of ACRL. For more from Steven visit his blogs, Kept-Up Academic Librarian, ACRLog and Designing Better Libraries or visit his website.

Create a Maker Program in Your Library
School Library Journal’s newest installment of Maker Workshop will feature up-to-the-minute content to help you develop a rich maker program for your library. During this 4-week online course, you’ll hear directly from expert keynote speakers doing inspiring work that you can emulate, regardless of your library’s size or budget. Course sessions will explore culturally relevant making and how to assess your community’s needs, mobile maker spaces, multi-media, and more!
Design Institute Heads to Washington!
On Friday, October 20, in partnership with Fort Vancouver Regional Library—at its award-winning Vancouver Community Library (WA)—the newest installment of Library Journal’s building and design event will provide ideas and inspiration for renovating, retrofitting, or re-building your library, no matter your budget!


  1. Maureen O'Brien says:

    An interesting and thought provoking read. I highly doubt that AI technology can replace librarians in the area of having high level conversations with PhD students and academic researcher in the areas of predatory publishing, publishing and author impact strategies and discuss ethics. I suspect even if it were possible, it would not be in my career lifetime. Complex algorithms and all that AI entails cannot replace the ‘human touch’ or human discernment.

    • You are asking the big question Maureen. What is unique, creative, interpretive, etc. about the work of a research librarian that would be mostly impossible for AI to manage. I think that gets back to Dan Pink’s point about how to be in an irreplaceable career. I tend to agree this would be hard to do – but hard to convincingly say it’s impossible – especially if AI has some type of learning capacity in which it grows the ability to “learn” how to interact with, say, doctoral students. As you say, it may be beyond our time.

  2. The future of librarianship is surely guiding patrons as they seek to navigate information. Suppose that robots end up being able to answer basic reference questions and provide tutorials on a myriad of information skills. There would still be an ongoing need to help our patrons navigate the information world beyond what robots can do, all those meta-operations that are represented in higher order thinking. Robots tend to do rote on a lot of things, not having the ability to exhibit expertise that helps people substitute better or best for OK. Librarians will just have to ramp up to the levels that robots can’t reach.

    • My reading about AI did suggest that it will eventually have the power of evaluation – so it could potentially help guide humans to the “best” information. Will it actually have the ability to help people navigate information? Hard to say. We’ll have to see where this goes, but I think you may be right. Thanks for your comment.

  3. A few key features that a robot will perform well over a human are: speed, ubiquity, self learning and 24×7. Currently machines can be programmed to learn by themselves and at the rate of such technology development (i.e. Deep Learning), my guess is that within 50 years the number of librarians required will be much less. On Youtube, real time speech to text translation is a reality and Deep Learning can be applied in almost every field. Libraries can embrace such technology to build a virtual personal librarian for each of their patrons. For example, a virtual personal librarian can be dedicated to a young patron over the years until he or she becomes a senior. For the library administrator a gathering of such “virtual personal librarians” and understanding their reading preferences would certainly improve collection development and maintenance. Librarians will still be around and their productivity will skyrocket provided if they know how to make use of the robots.

    • Thanks for your insights Sung. This idea of a “virtual personal librarian” sounds like a way that librarians could provide an AI version of themselves – that could be available to provide some research help when a librarian isn’t available – such as 2:00 am when we are sleeping but students are awake and doing classwork.

  4. Thanks Steve, Drones, AI, Big Data, Deep Learning, Robitics & IoT will converge in the next few years. A successful virtual personal librarian will be the ones that can harness these technologies into a seamless service delivery and support.

  5. Nice to see we’re getting away from decrying the “technovandalism” that is digitization and acknowledging that our potential robot overlords might not be all that bad…

  6. Barbara A Wood, MLIS says:

    Thank you for bringing this issue to the fore. The obsolescence of librarianship as a profession is the elephant in the room. We as librarians refuse to address the eminent demise of our beloved profession. Isn’t it our duty to examine the ugly truth and share it with those hoping to enter a career for life?

    The level of denial within our profession has reached frantic proportions. The professional literature is awash with articles on ways to keep librarians employed and remain relevant to stakeholders. A review of recent library job descriptions illustrates we will do anything, or become anybody, to stay relevant; i.e, pretend we are patent and copyright attorneys (copyright and scholarly librarians), become in- house statisticians, (data librarians, cheminformaticists), and data managers of electronic health records (Medical Informaticists). In a real stretch, Librarians at the San Diego Public Library are trained to identify victims of sex trafficking. Frankly, I didn’t go into librarianship to become a social worker or a statistician, I find this denial of our true professional core to be very sad, if not pathetic.