Watson's next conquest: Business analytics
Last year, a computer named Watson (named after IBM founder Thomas J. Watson) beat the two greatest human Jeopardy! champions at the popular quiz show. Viewers tuned in to see Watson analyze a Jeopardy! clue, search through millions of documents written in human language, and return a correct answer – all in less than three seconds. It was a captivating television moment. But once the game was over and IBM engineers brought Watson back to the lab, one final question lingered. What did this new level of artificial intelligence mean for companies in the real world?
At its core, Watson is a computing system that can extract facts and understand the relationships in vast quantities of data with lightning speed. What sets Watson apart from other computing systems is its ability to understand natural, human language, which is inherently full of ambiguity. This power, combined with its ability to judge each possible answer in real time and decide which one is most likely the correct answer, marks a major leap in computing innovation.
This has enormous implications for business. Don Campbell, IBM Distinguished Engineer for Business Analytics, sees the potential. "In business analytics, we really need to enable businesses to get to the answer of a question, not just point to more resources," says Campbell. "There are all kinds of applications where being able to surface up an answer based on a fairly complex human question is very business critical."
Revolutionizing business analytics
Existing business analytics solutions available today recognize subtle trends and patterns in data to give insight that results in better business decisions. Business analytics has helped companies do everything from preventing high-value customers from leaving for a competitor, to up-selling to current customers, to helping develop successful products. With the daunting complexity of human language however, there are limits to what existing business analytics can do.
The complexity often lies in unstructured data, which is frequently in the form of emails, text messages, audio and video files and represents up to 80 percent of data within an organization. "You can imagine the data sources of today not just having the traditional structured capabilities that we've had in the past," says Campbell, "but adding all of this new content through social media and user generated content. That's a perfect place for Watson to add value and understand what the buzz is around your product from a marketing and customer relationship perspective."
The promise that Watson represents is a major breakthrough for business analytics. Mark Morton, IBM Cognos Express Marketing, envisions how Watson will help companies better understand their customers. "Imagine your call center is on the line with an unhappy customer," says Morton. "With access to things like Web 2.0 type thinking and social media, we might be able to identify that this person is not a large customer but is influential in social media. So perhaps you should make a more expensive offer to keep him happy because if he goes, he may take hundreds or thousands of Twitter followers with him."
Making confident decisions
Having confidence that you have the right answer, of course, is critical in business analytics. In years past, companies would have total control of the information in their database and could be very confident of the integrity of their data. Today, with the wealth of data outside of company's control, it's a different story. "We now have pollution of information that's coming at us from untrusted sources, but if we just ignore that information then we've lost any value it could bring as well," explains Campbell.
In situations where answers may not be black or white, Watson introduces the idea of the confidence element in decision making. "By applying algorithms to deal with the system's confidence in certain types of hypotheses and in possible answers, Watson provides a confidence level associated with the response," explains Campbell. "That is very valuable to a decision maker who is sitting on top of both internal data and external data which has a variable amount of trust."
This surely hits a sweet spot for industries like healthcare where answers can have life or death consequences. Campbell describes a scenario in the not too distant future in which "Watson can understand more about what the patient is describing and is also able to mine through all of the medical information underneath the covers to better come up with what that patient might be suffering from." And companies in other industries are lining up as well. Earlier this year, Citigroup announced that the bank will collaborate with IBM to explore how to best infuse Watson's deep content analytics into the business.
“In business analytics, we really need to enable businesses to get to the answer of a question, not just point to more resources.”
Watson for midsize businesses
With the increased number of instrumented and interconnected systems generating massive amounts of data, companies are looking to business analytics to make sense of it all. According to a recent IBM study, "Inside the Midmarket: A 2011 Perspective," 70 percent of midsize companies have plans to implement business analytics in their operations.
According to Morton however, Watson will not be the sole domain of large enterprises. In the future, smaller firms may be able to access Watson-like capabilities as well. Midsize firms, says Morton, "are the ones who are most in need of agility in response to fast changing conditions so something like this that's going to give them that extra half step in front of everyone else is very important to them."
In fact, to best meet the budget constraints of the midsize company, Watson may be commercialized as a highly scalable solution accessible via a cloud environment. Campbell predicts, "We could assemble the Watson pieces in a way that could answer business questions with a smaller amount of hardware, with a slower response rate that would still add value to a certain class of user."
Getting to the right answer
On a smarter planet, the demands of business will require the same kind of real-time response and advanced analytics that Watson offers. And similar to the Jeopardy! game show, a wrong answer in a business environment can have negative financial consequences. "In your business, you have those same kinds of risk parameters," says Campbell. "What confidence level do I have to have in order to change my pricing to a certain level? What confidence level do I have to have in order to give this specific drug to this patient? In the end, you don't want 10 answers. You want to have one right answer."
More from ForwardView
Join the conversation
SMBs real-world use of big data
Analytics: the real world use of Big Data