IBM’s Watson Clues Insurers In to the Power of Deep Analytics

Normal view (turn off text only mode)
You are here:

17 May 2012 | In The News

IBM’s Watson Clues Insurers In to the Power of Deep Analytics

Super-Computer makes appearance at Annual CEO Roundtable

A version of Watson, the computer famous for defeating a pair of human contestants last year on the long-running “Jeopardy!” game show, was a special guest at Munich Re’s 18th annual CEO Roundtable event for clients.

Over 100 guests were on hand for the two-day event held each spring to offer insight and perspective about key industry issues. Jamie Bisker, the insurance relationship manager for IBM Research and occasional Watson-handler, gave attendees some perspective on the technology behind Watson and its potential impact on the financial services. Later, guests had an opportunity to meet a portable edition of Watson in person, so to speak.


Bisker’s role as a research relationship manager for the insurance industry provides him the opportunity to get closer to financial services companies as IBM considers “use cases” for new technology. Appearances like the one at Munich Re’s CEO Roundtable give Bisker a chance to exchange ideas among business leaders for new products and applications for the technology behind Watson, called deep question answering, among other capabilities.

Understands natural language

In essence, Watson is a question answering system. Watson was built as part of IBM’s Deep QA project and created to rival a human’s ability to answer questions posed in natural language. So don’t think of Watson as a pumped-up search engine. Search engines match words or phrases submitted by the user with words contained in its databases and indexes of content available from the Internet, often returning tens of thousands of results for the user to sift through in order to find an answer.

IBM's Jamie Bisker demonstrates Watson.

Power of Deep Analytics

The Watson that appeared at the Roundtable was a more compact, demonstration version of its show biz cousin but still offered an opportunity for some guests to fulfill lifelong dreams of uttering that iconic phrase, “I’ll take ‘Literary Characters’ for $200” and test their own “Jeopardy!” skills against Watson’s.

Watson uses natural language processing to interpret the meaning and context of a question in order to answer the question, not simply give options for answers.

Computers easily understand structured queries and data (think: database) but not unstructured information like natural language. Consider this clue from the famous “Jeopardy!” match. Phrased in a format that’s typical for the game show, the clue itself is challenging even for a person to understand:

"The first person mentioned by name in “The Man in the Iron Mask” is this hero of a previous book by the same author.”

Simply put, Watson first analyzes the clue to determine what the question is asking for, then searches through hundreds of millions of documents within its knowledge domain to come up with thousands of possible answers. Watson then uses its knowledge of the relationship between words and its experience in answering similar types of questions to merge and ranks its answers. All of this happens in seconds.

Beyond “Jeopardy!”

In August of 2011, health insurer WellPoint announced its plans to use Watson-based solutions to improve patient care via evidence-based diagnostics and treatment recommendations. This effort stems from prototype work done in conjunction with Columbia University Medical Center and several other academic institutions soon after the “Jeopardy!” match early last year. More recently, Citigroup announced in March of this year that it has an agreement with IBM to explore the possible uses of Watson technology to improve customer services for banking customers. In these cases and other strategic ventures Watson is engaged in, Bisker emphasized that Watson’s “jobs” represent a collaborative process.

“We don’t see Watson as an off-the-shelf product. We want a say in how it’s used,“ he stressed. “We’re looking for the game-changing – the world-changing – app.”

What are the possibilities for property-casualty insurers? A Watson-like question answering system could advance self-service by answering agents’ questions about products or customers’ questions about policies. Or it could automate accurate answers to complex claims questions posed by policyholders or even third-party claimants. Each answer would add to the body of knowledge, making the system smarter. Both applications free up call center staff from mundane and error-prone tasks, enabling a company to provide  better service to a greater number of customers.

Paired with IBM’s Cognos financial reporting software and SPSS statistical analysis software, Bisker says, a Watson-like system could augment professional judgment and free people up to “ask better, more strategic questions.”

In “Jeopardy!”, Watson’s answer “D’Artagnon” helped Watson eventually win $1 million dollars that IBM donated to charity. The same technology that helped Watson answer a “Jeopardy!” question could have enormous implications for healthcare, finance or virtually any other industry where people use information to help make better decisions as IBM uses Watson to push the science of deep analytics and natural language processing forward. Ultimately, Watson’s success will be measured by its impact on society.

Main Navigation
Service Men

© Copyright 2015 Munich Reinsurance America, Inc. All Rights Reserved.


This publication is available exclusively to Munich Re clients. Please contact your Client Manager.