The Oracle at IBM

Brainstorming process improvement with Watson

Published: Monday, February 8, 2016 - 18:00

For those awake enough to respond, please supply the Jeopardy question to this answer: “A computer system that won a million dollars in 2011 with access to 200 million pages of content, including the full text of Wikipedia.”

If you thought, “What is Watson?” you’d be correct as far as the computer’s 2011 persona goes. But the quiz-show star has been hard at work for the last five years, earning medical credentials and paying back the gargantuan investment its developer has lavished on it. Given its accelerated learning curve in medicine, finances, law, and other subjects, the computer is a lot smarter today than it was in 2011.

When you can process 500 gigabytes per second, or the equivalent of a million books, you like to keep busy. Meet Watson the clinical diagnostician, Watson the C-suite advisor, Watson the financial soothsayer, and Watson the creative chef. 

Watson’s 2011 Jeopardy debut and win

Oh, and in its spare time (just a philosophical concept for an immortal set of processors and algorithms), Watson is picking up conversational tips from students at Georgia Tech, whose mission is to teach the computer to “chat” about world problems with concerned humans. To do this, Watson first digested several hundred articles from the Public Library of Sciences website, and then pondered responses to the question, “How do you make a better desalination process for consuming sea water?”

“Seagull glands?” it offered. Apparently they can efficiently filter salt, which certainly puts them on the list for further research.

The chat project isn’t just drawing-room whimsy. Despite their inconceivable hoards of data, question-answering computers often have trouble with the linguistic shortcuts and blind alleys that human brains routinely traverse. It doesn’t help that humans use other signals besides words to communicate. But as a specialist in natural language processing and machine learning, Watson improves and refines its responses, and does so in fairly short order. 

Chef Watson

According to IBM engineer Steve Abrams, Watson’s foray into culinary matters has been a “metaphor” for the sort of creative thinking it’s been built to do: absorb a bunch of cookbooks, add a few spreadsheets on the molecular makeup of flavor and odor compounds, stir in a dash of food theory, run it all through self-correcting algorithms, and out comes Cognitive Cooking With Chef Watson: Recipes for Innovation From IBM and the Institute for Culinary Education (Sourcebooks, 2015).

Watson in the kitchen at the Institute for Culinary Education

To say that some of the cookbook’s offerings stray well into the outfield of food fusion—the Austrian Chocolate Burrito with edamame and apricots earned a “worst burrito I ever had” from one foodie—is to say nice try, especially for a chef that can’t smell or taste. No doubt Watson has already learned that cheesecake with mushroom-flavored whipped cream might jive with the best chemical parameters but still leave humans distinctly underwhelmed.

Big daddy of big data

However, Watson’s handlers are confident of its lucrative future as a cognitive helpmeet. In 2014 IBM invested an additional $1 billion in the technology and created the Watson Group, a think tank of some 2,000 professionals based in New York, with another scheduled to open in San Francisco this year. The division “will focus on R&D and commercialization of software, services, and apps that think, improve by learning, and discover answers to complex questions by analyzing massive amounts of Big Data,” as the company’s press release related.

Much of this will be cloud-based, which will give Watson and spinoffs great latitude as consultants. But for acquisitive corporations with several million dollars to spare, the computer has slimmed down from its earlier room-sized edition and is available in a sleek silhouette about the size of three stacked pizza cartons.

Watson’s intelligence is currently for sale in three service areas. It sounds like a travel agency, but the Watson Discovery Advisor helps researchers make useful connections among massive quantities of data. Watson Analytics crunches big data and spits it out in an interactive format for our increasingly overwhelmed work brains. And Watson Explorer offers “a contextually relevant view of any topic for business users, data scientists, and a variety of targeted business functions.”

The doctor is in… the cloud

In the medical arena, the computer is being pitched as a sort of noncompetitive colleague. The idea is to use it to find personalized treatments for patients, and for now the focus is on cancer.

To that end, Watson joined physicians at the MD Anderson Cancer Center at the University of Texas, where for the first months of its fellowship, accommodating humans fed it names, ages, genders, medications, lab tests, imaging results, and notes about thousands of leukemia patients. Then doctors taught it by entering accepted courses of treatments, and the reasons why they were considered so. Then Watson read a lot more medical journals. Finally, it was ready to brainstorm treatments.

Watson at med school

“When we first started, he was like a little baby,” Tapan M. Kadia, an assistant professor in the leukemia department, told The Washington Post’s Ariana Eunjung Cha. “You would put in a diagnosis, and he would return a random treatment.”

But “he” has improved because that’s what Watson was designed to do, and although the computer isn’t expected to replace the demigods of diagnosis any time soon, it’s hoped that computer systems such as this will improve the quality of care for a great number of people. For one thing, the expertise would be available not just in the major city where the human expert lives and works, but anywhere there’s Internet access.

Dance with me

Watson critics tend to point their fingers at the company behind the machine, citing IBM’s reputation for proprietary secrecy. “Why pay millions for Watson when you can run Hadoop for free?” wonders Matt Asay of readwrite, referring to the open-source software framework that stores massive amounts of data while offering enormous processing power for countless concurrent tasks. Hadoop helped create Watson, and it’s the software’s iterative potential that allows developers to fashion and fine-tune systems into useful innovations. “Watson, for all its technical marvels, largely ignores the very developers that could make it popular and useful,” says Asay.

At last month’s Consumer Electronics Show (CES), IBM hastened to its own defense. “It’s the dawn of a new era,” proclaimed IBM CEO Ginni Rometty. “The cognitive era.” Transparency is evidently an accepted element of it, since the company is establishing a platform and set of application programming interfaces. Companies from 17 different industries are now building solutions on top of Watson, according to Stephen Gold, the Watson Group’s chief marketing officer.

IBM is also wooing another hard-sell segment—startups and small businesses that have neither the resources nor connections of governments and giant corporations, the company’s usual partners. IBM has invested $100 million to create relationships with startups, and in the hail-fellow atmosphere of the CES also made noises about helping them connect with some of the big players. More than 500 startups are now participating in IBM’s cognitive era.

Clearly, IBM has read the tea leaves, no doubt with Watson’s help, and concluded that a collaborative and flexible strategy in the coming years is in its best interests. Watson and systems like it are powerful vessels for navigating the high seas of limitless information, and IBM recognizes itself as an early trader on those seas, making first contact with a future whose geography is only now coming into focus.

The company’s power and general optimism can be inferred in one of its recent ads about Watson. Somehow it coaxed music’s crankiest icon to exchange a few words with humankind’s next best friend. Where Bob Dylan leads, can the rest of us be far behind?

About The Author

Taran March @ Quality Digest

Taran March is Quality Digest’s editorial director A 35-year veteran of publishing, she has written and edited for newspapers, magazines, book publishers, and universities. When not plotting the course of QD with the team, she can usually be found clicking around the internet in search of news and clues to the human condition.