IBM’s Watson BRINGS Cognitive Computing TO CUSTOMER ENGAGEMENT

While the technology is still new, early developments are piquing interest in its future potential By Leonard Klie

IBM proved to the world that it was able to overcome a grand challenge when, in 2011, its speech-enabled Watson supercomputer summarily trounced several champions on the TV game show Jeopardy! Watson's victory demonstrated that computers could quickly process a vast amount of natural language content with a high degree of certainty.

Not surprisingly, businesses took notice and in May 2013, when IBM unveiled its Watson Engagement Advisor for customer interactions, five early adopters came on board. Among them were ANZ Banking Group and consumer research and ratings firm Nielsen, both of which signed on to explore how the IBM Watson Engagement Advisor could help enhance service to their customers.

A key tool of the Watson Engagement Advisor is Ask Watson, which lets customers speak directly to the technology through instant message, text message, email, Web chat, or a dedicated app on their mobile phones. IBM's Watson Engagement Advisor provides answers to questions that would otherwise have been fielded by a human operator.

At Nielsen, the Watson app will help advise those responsible for buying ads based on its ratings, letting them ask questions about how the company compiles its results or advising them on how to best spend their advertising dollars. The company also plans to apply the Watson technology to improve measurement of ad effectiveness and media planning.

For ANZ, Watson will be used to help customers with insurance plans, telling them where they might have too much coverage or not enough, based on an analysis of their policies. It also will be helping investment counselors tailor their offerings based on specific customer needs.

The other early adopters were Malaysian telecommunications provider Celcom, Royal Bank of Canada, and IT services provider IHS.

Analysts have also been impressed with Watson. Because Watson can listen and respond to a series of questions-including follow-up questions-and remember and learn from previous questions that were posed, Frost & Sullivan awarded the IBM Watson Engagement Advisor its 2013 North American Award for New Product Innovation in September. Gartner also cited Watson in its Top 10 Strategic Technology Trends for 2014 and predicts that by 2017,10 percent of all computers will be able to learn from past interactions as Watson does.

Watson's Hurdles

Despite this recognition, the technology still has several hurdles to overcome. Currently, the Watson Engagement Advisor remains untested by a broad consumer audience, and reviews are mixed as to whether it's up to the customer service challenge, especially given its limited speech capabilities.

"IBM Watson is a revolutionary natural language understanding engine with sophisticated abilities for processing language and learning from existing texts," notes Deborah Dahl, principal at Conversational Technologies and chair of the World Wide Web Consortium'sMultimodal Interaction Working Group. "It's likely to be a key component of future conversational systems with humanlike interaction abilities. However, several capabilities need to be added before Watson can be effectively applied to spoken interaction systems."

Technology writer and futurist Paul Gilster notes that while Watson's potential in customer service "is vast," it's likely that two or three more system upgrades are needed for the speech capabilities to be robust enough to live up to Watson's early hype. "I'm mostly looking toward the next-generation Watson and the one after that, when speech recognition should be totally seamless," he says.

Dan Miller, founder and principal analyst at Opus Research, says IBM is uniquely positioned in this regard because of an ongoing partnership with Nuance Communications. "IBM, with its partner Nuance, is up to tackling the challenges of speech-enabled customer service," he says. "Automatic speech recognition technology today is certainly accurate enough to serve as the grist for high-quality customer service, but it's not something that IBM, even with Watson, will do alone."

Nonetheless, Miller is confident in IBM's ability to master the intricacies of such an application. "IBM is one of a handful of companies that can dominate customer engagement technologies," he says. "Its long-standing investment in Watson has led to great advances in natural language understanding and machine learning."

That is where Watson truly shines. W atson proved on Jeopardy! that it has a unique ability to understand the nuances of human language, process questions asked in a format that mirrors the way people think, and quickly cull through vast amounts of data for relevant, evidence-based responses to questions thrown at it. The beauty of Watson, IBM claims, is that its cognitive systems can understand the context within users' questions and even improve its own performance by continuously learning from experiences.

Consumers can interact with Watson in plain English, directly or through an agent, to get personalized responses to questions and receive valuable insight with supporting evidence and a confidence score. The integrated Ask Watson feature greets and offers help to customers via a Web site chat window or a mobile push alert. "Speech recognition is just a shell for Watson," Miller says, "because the power of the Watson platform revolves around deep domain knowledge with natural language processing and machine learning."

Watson, in its current form, "is basically a way to get at the implications of big data," Bill Meisel, executive director of the Applied Voice Input/Output Society (AVIOS) and principal at TMA Associates, stated in an email to Speech Technology magazine.

"An organization might have a number of text-based resources with information that might provide an answer to a customer's questions. Watson analyzes this body of data and converts it into answers accessible by natural language inquiries," Meisel explained. "The focus is inquiries by text...although any speech-to-text engine could create a text entry in theory."

IBM has said that it hopes to eventually equip Watson with a conversational speech interface similar to Apple's Siri. A member of the IBM Research team in India working on the Watson project confirmed to Speech Technology magazine that "speech technology integration with commercial Watson offerings is planned for the future."

But for Watson to truly succeed in the customer service role, it will need to be expanded with the ability to carry on a conversation, Dahl says. The current Watson Engagement Advisor lacks the ability to collaborate with the customer on resolving an issue, taking an order, helping shop for a product, or troubleshooting a problem, she states. "In those kinds of situations, Watson will need to be able to proactively ask questions," she says. "Right now, Watson's focused on in-depth answers to specific, one-off questions that don't relate to each other."

Also critical to Watson's success as a customer service tool will be its ability to link to other systems and data sources. Dahl says developers will need to be able to connect Watson's language understanding abilities to business rules and enterprise-level structured information. "This," she points out, "will allow Watson to do things like find and talk about customer-specific data, including addresses, purchase history, and preferred-customer status."

Eventually, IBM expects the Watson Engagement Advisor to be used for additional purposes, such as helping customers book travel plans. "Customer engagement is a natural fit for Watson, which can instantly create a strong bond between who customers are as individuals and what types of information will help them reach their goals," Manoj Saxena, general manager of IBM Watson Solutions, said in a statement.

Developing with Watson

Since Watson's appearance on Jeopardy!, IBM claims to have improved its system performance by 240 percent and reduced the physical requirements needed to run it by 75 percent. Watson reportedly can be put down on a single server today, opening up a whole new world of opportunity on how and where it can be put to work.

Additionally, IBM has revealed more than a dozen partners across a number of industries with whom it is working on developing new big data a nasties programs using the Watson technology.

Other industries finding uses for the Watson computing platform have included the medical field, with Memorial Sloan-Kettering Cancer Center in New York, the University of Texas' M.D. Anderson Cancer Center, Cedars-Sinai Medical Center in Los Angeles, and several other top hospitals fielding it to help doctors design the best course of cancer treatment based on the latest clinical research and patient data. The product being used, Interactive Care Insights for Oncology, developed through collaboration between IBM and Sloan-Kettering, is expected to launch as a commercial product for hospitals and cancer care providers later this year.

In the insurance sector, IBM has worked with WellPoint to create an app that looks at a patient's medical history prior to processing a medical claim to determine if the treatment request is in line with company policy. And in financial services, IBM is working with banking firm Citi to deploy Watson as a financial adviser.

For other organizations looking to leverage Watson's capabilities, IBM in November launched the Watson Developers' Cloud, designed to help entrepreneurial software developers create new applications infused with Watson's cognitive computing intelligence. The Developers' Cloud includes a developer toolkit, educational materials, and access to Watson's application programming interface.

Fluid, an Internet start-up that helps brands develop personalized shopping experiences, later this year hopes to release the Expert Personal Shopper (XPS) app powered by IBM Watson to offer consumers assistance in their purchase decisions. The app is expected to draw data from underlying sources, such as product information, customer loyalty data, sales histories, user reviews, blogs, and relevant magazines and other publications, to give users a highly enriched and personalized shopping experience. Fluid XPS, which the company hopes will be able to accept voice or text inputs, will be able to have conversations with customers and then recommend clothing based on their tastes.

Fluid has been experimenting with a prototype of the XPS app for The North Face. Called the Compass Gear Advisor, the app acts like an experienced in-store salesperson to help customers find the right coat or outdoor apparel for their particular needs.

MD Buyline, a Dallas-based medical supply company, plans to launch Hippocrates, a Watson-powered app that will help doctors make more informed medical equipment purchases. Welltok, a Denver-based social health management company, is bringing Watson technology to its collaborative app CaféWell Concierge to help consumers craft personal healthcare programs based on their medical conditions and available information related to those conditions.

Showing its commitment to the development community, in early January, IBM launched the IBM Watson Group, a business unit dedicated to the development and commercialization of cognitive computing innovations delivered through the cloud. As part of this initiative, IBM allocated $1 billion to the IBM Watson Group. This will include more than $100 million available for venture investments to support IBM's ecosystem of start-ups and businesses building new Watson applications in the IBM Watson Developers' Cloud.

Looking Ahead

IBM CEO Ginni Rometty in October said publicly that two new versions of Watson are in the pipeline. Watson 2.0 will be able to scan pictures, including X rays, and interpret them. Watson 3.0 will have the ability to debate and reason.

"The current technology already handles all types of queries that customers of an enterprise ask of the enterprise," the IBM Research team member in India said in an email. "We are definitely already building Watson systems for self-service that are automatically interacting with humans. We are, in parallel, advancing the state of the art of such interactions with more natural and intuitive interactions."

Dahl is optimistic about the future of IBM's Watson. "Augmented with speech recognition, conversational skills, and application integration capabilities, I believe Watson will become a central technology in future customer engagement applications," she says.

"IBM is one of a handful of companies that can dominate customer engagement technologies."

"Watson will need to be able to proactively ask questions."

News Editor Leonard Klie can be reached at [email protected].