Watson: Coming to a Cloud Near You
The cognitive computer known as Watson will soon be available as a service in the cloud. For now, we’ll resist calling this “WaaS” and opt instead to point it as another step along the path of creating more robust, big data-ready cloud services.
“We’re now starting to accelerate the development of our platform – so we’ll be able to offer the Watson technology as a service,” Stephen Gold, a director with IBM Watson solutions, shared with Datanami this week. Gold revealed that IBM has begun an internal pilot and the company is currently working on ways to infuse the Watson technology into its entire portfolio.
“What’s happening right now is that we’re actually piloting this internally,” said Gold. “We’re offering this service to our other brands and products and they’re starting to incorporate the technology into their offerings. We’ll use that, if you will, as our beta site, and then later this year we’ll take a handful of our partners and work through the system of how to deliver it.”
From there, says Gold, IBM will begin to build and accelerate on moving the Watson offering into the broader market, making the technology available as a service through the cloud.
In case you aren’t familiar with Watson, it’s what IBM calls a “cognitive” computer – a supercomputer system that can understand questions posed in natural language, and deliver results based on whatever data is in its vast stores to access. Watson made headlines in early 2011 when it was showcased on the game show, Jeopardy, in a Paul Bunyan-style showdown against two of the greatest Jeopardy champs in the game show’s history. In the man vs. Machine matchup, Jeopardy champs Ken Jennings, and Brad Rutter were no match for Watson.
Two years later, IBM tells us that Watson has made great strides. “Watson has gotten really good at understanding content and context and navigating complexities of speech,” Gold told Datanami. “When you ask a question, it looks at the semantic interpretation – it parses it and looks for candidate answers. A simple question that I may ask, that you may ask, that others may ask – we all express it differently. Watson’s ability to understand not just the question, or the query, but extensive input like a patient file opens up a whole new world of possibility.”
Gold likens big data to fuel in Watson’s tank, making its abilities possible. Gold says that Watson ingests the natural language data on the front end that becomes the question, and then cross-references it against a vast repository of knowledge in its big data stores and then retrieves back with confidence what it believes are the appropriate set of responses.
Gold is careful to say that Watson is not finding “answers,” per se. “What’s a little bit different is that Watson isn’t looking for an answer,” explained Gold. “In fact, it’s not very good at finding answers – it’s really good at finding its confidence in a series of responses. So when you think about big data and you think about the implications – someone asks a question, you want to sort through big data, you want to find the answer, but in a lot of cases, it’s a whole set of answers.” Watson, says Gold, is able to bring back with confidence a set of weighted responses.
If that sounds an awful lot like a search engine to you, you’re not that far off, explained Gold. “What Google is to search, Watson is to discovery,” says Gold. “Search is synonymous with the idea that I put in a couple of keywords and it very quickly retrieves massive amounts of matches. But the individual has to sort through each of the retrievals and try to find the information. With Watson, it’s always going to try to put the question in proper context and only bring back what’s relevant.”
Currently, Watson is being used in industry in limited applications. Gold says that they now have about a half dozen use cases that are in different stages of development. “The first two production instances are in healthcare,” said Gold. “We’re moving into other industries. We’ve announced financial services and are now looking at a product that we’ll announce in the May timeframe that will take us cross industry.”
In the meantime, IBM is doing what it can to encourage people to be open minded about the applications Watson might have in the market at large. To that end, they have taken the challenge for use cases to the next generation through a competition called the IBM Watson Academic Case Competition. Through the case competition, IBM challenged 100 students at the University of Southern California (USC) to create business plans for applying Watson to pressing business and societal challenges.
The students were given a crash course on the Watson technology, including a demonstration of how Watson is currently being used to help WellPoint, Inc. and Memorial Sloan Kettering Cancer center improve the speed and quality of treatment for cancer patients. The students were split as a composite across the colleges – engineers mixed with business and liberal arts majors, and each of the 24 teams were given 48 hours to come up with a business plan to present to a panel of eight industry and faculty judges.
The winning plan was a theoretical Watson implementation in legal research. The first place team reasoned that Watson’s natural language abilities lend itself well to the legal field, where users would be able to use Watson’s big data processing power to sift through online legal documents for facts, identify evidence to support a case, and forecast the probability of success. Other placing entries included the use of Watson to optimize employee training, and the use of Watson by physicians to identify returning soldiers who may develop PTSD by uncovering insights from the vast amount of data in military records.
While these theoretical implementations may be a ways off, IBM says they are bringing them closer to reality as they prepare for their Watson-as-a-service cloud offering, and look forward to seeing how Watson gets used to augment systems that are already in place, and shift computing in a fundamentally new direction.
“What we see in cognitive computing, cognitive systems are going to extend and compliment what’s already there,” says Gold. Cognitive computing won’t be disruptive in the sense that it will replace systems, explained Gold, but rather is about how the traditional systems can be extended and complimented to amplify the capabilities that organizations have.
“I think the center of cognitive systems is the individual who benefits in whatever capacity they’re touched, and I think that’s a big deal.”