Your Call Is Very Important to Us

If you watch the TV quiz show Jeopardy!, you probably saw a computer named Watson go where no machine has gone before: winning a recent match against Jeopardy! champions.

IBM's Watson is something new: a computing system that analyzes human language and provides specific answers to complex questions at rapid speeds. On Jeopardy!, Watson time and again correctly answered the kind of subtle, difficult questions that the quiz show is famous for. An example, from a Jeopardy! category called Literary Character APB:

Host Alex Trebek: Wanted for a 12-year crime spree of eating King Hrothgar's warriors; Officer Beowulf has been assigned to the case.

Watson: Who is Grendel? [the correct answer]

Watson represents a tremendous breakthrough in the ability of computers to understand natural language that humans use to capture and communicate knowledge as opposed to specially designed or encoded language just for computers. It can evaluate the equivalent of hundreds of millions of pages of material–books, reports, articles, and so on–in three seconds or less. It is not stymied by intricate wordplay. In the above example, modern crime lingo was used to pose a question about a literary classic (Beowulf) from 1,000 years ago. Watson was unphased.

Think of the potential uses of this amazing technology. Watson could serve as the foundation of a digital medical assistant, helping doctors diagnose diseases and make treatment recommendations in real time. It could give bankers the ability to get answers quickly to incredibly difficult questions about strategic decisions and market changes. 

Watson, named after IBM's founder Thomas J. Watson, could even make our everyday lives easier. For example, it could eliminate the frustrating automated customer service experiences that have plagued the world for decades, wasting countless hours, fraying nerves, even damaging customer loyalty.

We've all used these automated phone systems. Push 1 if you want X, push 2 if you want Y, and so on. Making a selection just leads to another question and then another, and before you know it, you've wasted 15 minutes pushing buttons. You can't press 0 to get an operator. The system won't let you.

Watson technology can turn these excruciating encounters into a pleasure. In a retail setting, Watson would be tied into all of the company's product and customer information and could answer the most difficult questions immediately: How do I connect the Blue-Ray to the TV? What about the Internet? Do I use the white cables or the blue cables? How do I install the security software on my laptop? What's wrong with my cell phone?

Watson technology would be just as useful when trying to get an answer from the government, whether it's dealing with the local motor vehicle department or a big agency of the federal government. These entities (even the small ones) just have too much data and no good way to get at it. So they leave it up to you. The endeavor usually results in a journey through a labyrinthine tangle of websites that would horrify Kafka.

Watson technology could provide immediate answers to citizens' questions on a broad spectrum of topics. What are the zoning regulations for building a new porch? Does the Alternative Minimum Tax apply to me? What's the best way to get a visa?

At the heart of this new capability is Watson's facility with language as it is spoken by real people; its skill at plucking the exact, correct answer from an ocean of unstructured data; and its ability to learn.

In the history-making Jeopardy! match, we can glimpse the seeds of something genuinely new. Watson technology is poised to vastly improve the experience that people have with large institutions, both public and private. The potential benefits are breathtaking–a better informed, more engaged citizenry, and a dramatic increase in the ability of everyday people to access the information essential to modern life. 

Carolyn Heller Baird (cbaird@us.ibm.com) is the global CRM research leader for IBM's Institute for Business Value.

CRM Covers
for qualified subscribers
Subscribe Now Current Issue Past Issues