At the Only Edge that Means Anything / How We Understand What We Do
by Dennis Brunning (Director, The Design School Library, Arizona State University)
In the Knowledge Factory — My
Shift at IBM Watson
Well, to be honest, half a shift. And not so much a factory floor but four hours of grueling seat time, display, talking heads, watching Watson triumph over man and men, and other stuff of a tight training day.
It’s been about five years since that eventful day in cold, cold Poughkeepsie, New York, when IBM engineers built a makeshift television studio in an IBM lab. On January 11th, 2011 Big Blue hosted Merv Griffith’s Jeopardy. It was not just any taping session; two of the long running game show’s big winners, Ken Jennings and Brad Rutter, joined Big Blue’s heir, Watson, to contest Watson in three televised episodes. Watson, although stumped by the non-fiction category and unable to bet well in daily doubles, won handily by over $50,000.
Watson was IBM’s self-challenge, to build a machine that could outthink humans. In 2005 Deep Blue, an early version of Watson, handily check-mated the world chess champion, Garry Kasperov. This chess playing mainframe had been fed the text data of thousands of chess matches and all the documented strategies. Deep Blue was Bobby Fischer without the attitude.
Jeopardy was a no brainer for the ultimate brain off between man and machine. For over 20 seasons, three contestants had battled each other in a fast-paced, quick-witted buzz, bet-your-money TV version of trivial pursuit. Who better to prove a point that computers can think but to challenge the best thinkers in television broadcasting game show history?
But the pounding in Poughkeepsie was years ago. Those of us gathered at Phoenix’s IBM Training Center early in 2015 were there spurred on not only by our own appreciation of Watson’s triumph over us mere humans but also by curiosity: how was Watson at making everyday money?
What did we learn IBM Watson ROI for the enterprise?
First our hosts surveyed us about what motivated our attendance. Why were we there? Over 90 percent of the thirty or so trainees thought and felt that Watson had great relevance to their jobs and companies. Seven percent were moderately interested, and a mere three percent didn’t know why they were there. Their bosses said go.
The IBM Watson people were direct and clear about the Jeopardy challenge. The challenge was a proof of concept, whether a computer could beat humans in the real world. They pointed out, though, that Jeopardy Watson was specifically built for Jeopardy. The huge database, sourced with libraries full of reference works, all Wikipedia, and immense slices of the factual Web and more, were dumped into memory to compete with two humans each with three pounds of brains. The IBM folks even sheepishly admitted that Watson had trained in over 100 matches with IBM personnel with questions and answers mined from all episodes of the longest running game show in American television broadcasting.
Questions? Was this, well, sort of a setup? The Jeopardy match? Yes. Would many of us have the same resources as Jeopardy Watson? No. Or maybe. Probably no.
What then, would we have and what would we get?
Our training centered on the real world 2015 business application of Watson. We learned of the Watson Knowledgebase which included a rich array of business reference and data sources. We learned about its natural language processor that would work with leading voice recognition systems to parse regular queries into terms Watson could recognize and manipulate. We learned about Watson’s ability to emulate decision making, its algorithms that emulated the best practices thinking of world class enterprises.
All this would not let you take Watson home or even dial it up and play chess or engage in intellectual discussion. What you’d have would be a smart interface between you, your data, and Watson’s analysis of the data.
Our trainers used a Watson investment firm application. On our screen was a dashboard. Imagine your investment adviser, the person in charge of your retirement. He or she must handle streams of real time data, a profile of what you want, what you have, and where this puts you twenty years out. The investment world has this but as they say living in data silos.
Watson can help organize this for your investment adviser, alert when actions need be taken on yours or similar accounts, all the while suggesting to you, with the help of algorithms drawn from all the data, new customers whose money and data could add to the Watson’s knowledge. More the merrier…
Nifty right? We trainees thought so. Remarkably, too, IBM has nicely priced options — even free ones — that help introduce us and our organizations to Watson’s advantages. There are development kits and Websites where we can play with running our data against Watson and build our own dashboards. We can add Watson to our payroll as a librarian/researcher and consultant/know-it-all.
Surprisingly, our exit survey revealed only about 60% of us are convinced now that we should or needed to be at Watson training. I think we were curious about just registering with Watson and starting to roll out a form of Watson in our workplaces.
At break some of us dissed on a common problem with the Watson outlook. There was a little of that engineering hubris that demands a built solution to life’s problems. Then there was also what we joked was Watsonhausen Syndrome by Proxy. Watson’s exaggerated claims for recognition as a huge step forward for the machine mind when its use would be more like teachable full-time assistant.
Although it’s astonishing technology, we were less enthusiastic leaving than entering. The knee-jerk response, especially among public sector employees, is that we could not afford this. Even if we fired everyone, we’d be a day late and a dollar short. And with everyone laid off, how would Watson easily learn what it needed to learn to replace us?
We’re rounding the fourth turn at the 2015 race and IBM has stepped up marketing for what Watson can presumably do. Watson beating the Jeopardy twins was a billion dollar proof of concept exercise, a reverse loss leader to get the conversation going. The training session was just that — to learn to think another way, the Watson way.
As we shuffled out, our mutual looks spoke to a larger realization. Watson represented a truth and reality bright on our human horizons that signaled the transformative moment of machines off-loading rote decision making and factual drudgery to their CPUs and allowing us humans to do something else. We left, equally light and heavyhearted. Watson knew us better than we knew ourselves. As Jennings’s joked in Final Jeopardy, “I for one welcome our new computer overlords.”
And it really doesn’t matter if
I’m wrong I’m right
Where I belong I’m right
Where I belong
See the people standing there
Who disagree and never win
And wonder why they don’t get in my door
— Paul McCartney
My wife Cathy knows I’ve got an Amazon habit. We buy everything from Amazon because it is so easy. For my own good and our retirement money she’s wise to ask about any uptick in Amazon boxes piling up daily at our front door.
Cathy orders from Amazon too as well as Kohl’s, Etsy, Target — it’s a long list. She’s retired and home though, and gets her packages stowed away before I get back from the office. If I’m lucky, she’ll neatly stack my boxes in my little office off the front door. If unlucky, boxes will be strewn around the porch or stacked like a Leaning Tower of Pisa next to a planter where we grow dead plants.
It’s unlucky for me because it’s an iconic statement of my excess. All those Amazon boxes, each one of them charged to our overheated Amex card. Every one of them, no doubt, a conspicuous consumption. Not in Veblen’s sense of keeping up with the Joneses’ type behavioral economics. No, more precisely, Dennis’ obsession with easily ordered and purchased Amazon books.
Yes, I confess my wife is right and so are you, dear reader, if you sense how wrong this may or could be.
There is method in my madness. It’s obsessive, yes, to Amazon One-Click for books. But how can I, a mere librarian, resist buying the library books that were on my reading as a kid?
Yes, I’ve discovered as perhaps some of you have that our library books are going for as little as one cent a book. Of course, shipping, handling, and taxes add another three or four dollars. And some used bookstores don’t figure into Amazon Prime which in the Amazon used trade business doesn’t save you shipping but does speed up shipping.
So I’m buying back a library, my library, from your libraries, book by book. The books show library wear and my shelves now seem, to the noticing eye, as lifted from a public library here, a school library there, a defunct school of higher education.
I just received a great library rebound copy of John Cheever’s exquisite short story collection, Some People, Places, and Things that Will not Appear in My Next Novel. This is not one of Cheever’s best or remunerative books. Yet it is memorable to me and as it has been overlooked by its publisher as a reissue; at three bucks, it’s a steal. In my home, it’s shelved alongside the Library of America’s edition of Cheever’s well-known novels — the Wapshot novels and Falconer. I love it as the library that tossed it didn’t. Love the penny price. Like a penny stock it has its own cheap charm.
Why? I won’t make fun of how we are dumping our intellectual property probably not even at fire sale prices. Besides the dumpers seem to be public libraries and school libraries — so read the leftover markings of ownership, call number, and date stamps. My penny book stocks from Amazon are faded, blurred, oddly marked as if weeded in haste or tossed because a well-worn book, is well…
My library suppliers are probably strapped for space and they’ve found space in the jettisoning of the Cheever’s, Bellow’s, and Updike’s. We’ll let Nicholson Baker find the humor and irony in this — although even Nick has moved on from libraries, librarians, and our shred, shrink-wrap, shirk, and high-density shelving behavior. Our loss is my gain. Sort of.