Watson-Type Automation: How Do I Get It?
Insurance Experts' Forum, February 24, 2011
In my last posting here, I talked about the recent Jeopardy beat down administered by Watson, an IBM-built expert knowledge computer, to two Jeopardy champions of recent years. IBM Global Insurance Industry Leader Jamie Bisker emphasized that the event was a demonstration of the increasing ability of computers to understand and work with natural language—an ability that would be useful in developing more advanced expert systems for insurance and financial services.
This caused me to wonder where an insurer or broker might get its hands on such technology, which would surely be a competitive advantage in terms of problem solving and product development. Unfortunately, such capability is not available off the shelf at your local computer outlet. According to Bisker, IBM is currently working on a Watson-style application for the medical field, which may be of interest to health insurers down the road.
“In three to five years we will be experimenting using our FOAK (first-of-a-kind) mechanism in insurance,” he says. IBM hopes to have a system that can provide text and other information on the fly to CSRs in an insurance call center based on what the computer hears from customers. Bisker points to current research in natural language interfaces being done at universities around the world. A Watson-type semantic engine, he notes, could provide small companies with information and responsiveness that are traditionally found only in larger companies, especially if the engine were cloud-based.
A Watson-style insurance system could helps claims reps who need immediate access to information in manuals to more quickly answer questions about history of accidents that are similar, adds Bisker. It could also “suggest” things to look into at the scene of an accident, based on the history of similar events.
While the technology holds much promise, however, the reality is that truly effective natural-language-based systems are still far away. To anyone who saw the Jeopardy competition, it appeared that Watson was responding to Alex Trebek’s “answers” as spoken. Actually, the “answers” had been pre-programmed electronically and were presented in that form to the computer at the same time as the human contestants heard them. So while some of us dreamed of a Star Trek-like computer that could actually respond to human speech, such was not the case here.
Another inhibiting factor on acquiring such a system was that Watson was far larger than a typical desktop appliance. In fact, it took up a whole server room at an IBM facility, meaning anyone who wanted to utilize such a system would have to have lots of spare room and lots of spare servers, not to mention lots of funding.
I agree that a Watson-like system for insurance would be a useful tool, but let’s temper our optimism with some practicality. We’re not going to be able to buy “Watson for Insurance” any time soon, but we can begin developing the expert systems capabilities that Watson promises right now. This must be a process that happens over time, however.
Another thought that occurred to me was how dangerous it could be for such a system to truly understand human speech. Say we programmed our system to defend itself against any intrusions or threats, which seems logical. What happens if some frustrated technician walks by and says, “I’d like to blow up that *&^%$# computer!” within earshot of our expert system?
I know. I’ve probably been watching too many science fiction movies.
Ara C. Trembly (www.aratremblytechnology.com) is the founder of Ara Trembly, The Tech Consultant, and a longtime observer of technology in insurance and financial services.
Readers are encouraged to respond to Ara using the “Add Your Comments” box below. He can also be reached at firstname.lastname@example.org.
This blog was exclusively written for Insurance Networking News. It may not be reposted or reused without permission from Insurance Networking News.
The opinions of bloggers on www.insurancenetworking.com do not necessarily reflect those of Insurance Networking News.
Add Your Comments...
If you have already registered to Insurance Networking News, please use the form below to login. When completed you will immeditely be directed to post a comment.
You must be registered to post a comment. Click here to register.