Gameshows, surgery, service...Where next for Watson?

31st Jul 2013

Deep Thought may have declared ‘42’ the meaning of life in The Hitchhikers Guide to the Galaxy, but for those of you not satisfied with Douglas Adams' hypothesis, there’s another supercomputer to put your existential questions to.  

IBM’s Watson is a cognitive system that can understand natural language and, in a move away from traditional computer systems, is taught rather than programmed.

Steve Gold, IBM’s VP of Watson Solutions, explained to that Watson is “a combination of computational linguistics, which we call natural language, and added machine learning, which has sometimes been referred to as AI. It’s taking Big Data and analytics and bringing it all together in a very unique prescriptive way and as a result, Watson is an embodiment of a new era of computing.”

Since rising to fame with the infamous Jeopardy! win against all-time champions Ken Jennings and Brad Rutter, IBM sent Watson to medical school to get a job as an oncologist. Using Watson’s ability to ingest and make sense of vast amounts of unstructured data such as periodicals, books, blogs and tweets, doctors at New York’s Memorial Sloan Kettering Cancer Center taught the computer to review oncological case histories, patient symptoms and hereditary histories to assist in diagnoses and treatments.

“Information in healthcare doubles every five years and the majority of information resides in unstructured formats like clinician notes or research studies or periodicals,” says Gold. “We want to be able to unlock that information that's relative to an individual case and then marry those together to get a better insight in terms of what logical action we should take and learn from that.

“Did the doctor follow the recommendation, did the patient show progressively better outcomes, and were the survival rates longer than what was historically estimated? Those are the things that become key data points for Watson to learn.”

During Watson’s work in healthcare last year, IBM discovered more of the technology’s capabilities and found the supercomputer embodies three very specific capabilities that mirror how we as individuals interact.

“It’s very good at the ask function, and that's a natural active dialogue of how I would use a Watson to interact through a series of Q&As. The second thing we discovered about Watson is that it's really good at discovering things so if you point it at large volumes of information, it would find things that would typically be difficult to decipher or to connect. And the third thing, it was really good at advanced decision making a helping individuals make very complex decisions.”

Uncovering these three capabilities led IBM to think about how Watson could be put to work in a commercial setting at a time when more and more companies were, and still are, focusing on taking the customer experience to a new level. Gold points to figures that show currently in North America, 270 billion calls come into the call centre but half go either unresolved or require escalation, which results in a lot of frustrated customers.

Putting Watson to work

In March earlier this year, IBM announced that it had struck a deal with Citigroup, allowing the banking group to use Watson’s technology to help improve the customer banking experience. Citi announced that it would be using Watson’s analytics capabilities, language processing and decision support to create a ‘first-of-a-kind customer interaction solution’ to help staff identify opportunities and evaluate risks.

More recently, IBM announced that it would be putting Watson to work in the world of customer service, using Watson’s as a customer service bot to replace automated customer service technologies. When customers ask a query via either speech or typed text, the supercomputer can then sort through a wealth of unstructured data to determine the best answer, and with the ability to learn as it goes, can keep track of customers’ likes and dislikes to deliver the most favourable solution according to each individual’s preferences.

IBM also introduced the ‘Ask Watson’ feature, which enables customers across multiple channels such as email, text and chat, to ask questions that Watson responds to in a unique and contextualised manner – rather than simply repeating past responses. Nielsen, Australian bank ANZ and Royal Bank of Canada were named as some of the first clients’ trialling Watson’s technology fir customer service.

“We recognised that Watson's capability to have this interactive dialogue could address the lion's share of those customer service situations and the more we talked to clients, like Nielsen, the more we validated the benefits of using this technology to really transform an experience,” says Gold. “As a consumer, much in the same way ATM's changed the way I got my cash, Watson has this ability to really change the experience by which I get problems resolved, questions answered or to learn more about a product or service that a vendor offers that I’m interested in.”

Gold added: “There’s opportunities to unlock insight into consumer preference, concerns, bias and so on, which can be put into context in order to make a decision about a product or service. Knowing how customers are pre-disposed towards certain benefits, features or functionalities or even if they want to deal with socially responsible organisations or sustainable products, all of that is discoverable opportunity in the social realm. It’s already happening, our clients are already mining that information, getting better insight, and understanding how to improve and respond to customers.”

Developing Watson’s technology in this way could also bring a new era of automated customer service that is sophisticated and accurate. As IBM’s Guy Stephens recently commented on “For me, Watson suddenly heralded the possibility that perhaps some of the questions that Esteban [Kolsky] asked around automation and scale can begin to be answered. I genuinely feel now, that we're actually about to embark on a new phase of growth, when the customer service model (and by extension the business model) itself will begin to be looked at."

So, having conquered Jeopardy!, excelled in oncology and become the world’s best customer service agent, where could Watson be put to work next? Earlier this year, analyst house Nucleus Research suggested that Watson’s capability to integrate unstructured information (such as emails, recorded voice conversations, twitter feeds, and other online sources of information) with structured data could better drive ROI within CRM projects – particularly those in sales management.

The report What Watson can do for CRM claimed that Watson could vastly expand the universe of data available to managers on how sales is or isn’t performing. “With this data, companies could make much quicker and methodical analysis on many processes that managers spend unnecessary time and effort on today, including incentive compensation, territory management and team assignment. Additionally, Watson’s ability to rapidly process information on sales conversations and related outcomes could result in the supercomputer proposing new responses,” predicted the report.

With the analyst house suggesting that 80% of CRM returns have not yet been achieved by the average company and 30% of those remaining return opportunities come from instrumentation and sales management, Watson could be "the ideal candidate" for solving many of the problems that CRM applications have yet to solve - sales management problems, says Nucleus. 

But other applications are also on the horizon - and Gold explains that when Watson first appeared on jeopardy there was a flood of ideas of where the technology could potentially be put to work.

“You'll continue to see Watson take on a general characteristic of working with situations where understanding information, time to make a decisions and accuracy of response are important, where the nature of interaction is such that it could be facilitated on this omnichannel 24/7 type always-on experience. Those are good dynamics that will define opportunities for but it's really going to be your clients that determine how Watson will be put to work.

“Much of the way we've gone into smarter commerce, it's our clients that are determining where in smarter commerce Watson will be used – be it on the service side, sales side, marketing and so on. I am absolutely astounded that every single day I learn a new way in which someone thinks about putting Watson to work,” he concludes.


Replies (0)

Please login or register to join the discussion.

There are currently no replies, be the first to post a reply.