At IBM’s Think Forum in New York, CEO Ginni Rometty took us through the success and future of Watson, IBM’s automated decision engine. This technology fascinates me because it’s the first major step to change the basic computing paradigm.
[ More: IBM Watson HQ Opens for Business in New York’s Silicon Alley ]
Until now, people have had to learn how to communicate with machines. With Watson, on the other hand, machines begin learning how to communicate with us. (IBM isn’t the only company chasing this; it’s joined by large superpowers such as Google as well as fascinating small companies such as Beyondcore.) We aren’t easy to communicate with, either, making this a difficult problem to solve, but the company that solves it will likely change computing forever, increasing the proven value from the related systems.
Events such as Think Forum show the next big technology revolution: The creation of thinking and learning machines that actually care what we want and can figure out how to give it to us – often despite our own biases and preconceptions – avoiding what would otherwise be horrid mistakes.
Treat Cancer, Communicate With Millennials, Inform Good Decisions, Book Honeymoons
One of the most compelling examples reviewed the case of a cancer patient. After reading a form that defined the patient’s condition, Watson laid out a course of cancer treatment with the highest probability of positive results – and also identified an area of concern that the doctor might have missed and, by doing so, put the patient at additional risk. Cancer hits home for me; anything that can improve care and reduce mistakes could very well save my life – or yours.
[ Related: Mayo Clinic Turns to IBM’s Watson to Fill Clinical Trials ]
The first panel, “The Business of Knowing,” then focused on communicating with millennials and their successors. On the stage were Rometty, Westpac Bank Gail Kelly, AT&T CEO Randall Stephenson and moderator Dr. Fareed Zakaria, who has his own CNN show.
Today’s customers frequently change their minds about the services they want and where they want them. Companies struggle to successfully communicate with them. Legacy systems can’t adjust to these changes as ever-younger generations come to market. If a company misses a wave, it misses an entire generation and, by so doing, effectively goes out of business. Decision engines seem particularly good at addressing this sort of problem – and the CEOs on the panel clearly recognize that their companies need this technology.
Another session talked about data as a natural resource. (I’ve argued for years that this should be the case, though it would mean the end of Google. It gets data for free, but governments tend to charge for natural resources that come from public lands – or citizens in this case). The panel included executives from Macy’s, IBM and BNSF Railway Company, but the most fascinating remarks came from Nate Silver, author of The Signal and The Noise.
Silver started with a statistic. Virtually all Republicans watch Fox News, while nearly all Democrats watch MSNBC. This means both groups see a vastly different world. (No wonder they can’t agree on anything.) He then pointed out that the information going into data analytics systems is often compromised by noise or bias. This lets people make bad decisions that seemed backed up by data but actually weren’t. As we rely more and more on intelligent systems, the adverse impact of compromised data driving bad recommendations will become a terminal-level problem for many companies.
The panel called “The Promise of an Intelligent Computer” got to the meat of the matter. Mike Rhodin, senior vice president of the Watson Group, opened by describing Watson’s competitive advantage. Done right, it will provide unbiased answers based on the massive amounts of information in the system.
[ Commentary: Big Data Success Is All in the Analysis ]
Dr. José Baselga, physician-in-chief at Memorial Sloan Kettering Cancer Center, provided more details on the cancer case mentioned above and argued that something like Watson is critical for a large hospital with thousands of patients and doctors who are spread thin. He drove this home by demonstrating how well Watson helped diagnose rare diseases and successfully identify procedures that could cure them.
Insurer USAA and DBS Bank also talked about using Watson to improve the customer experience. (I have home and automotive insurance through USAA, which constantly receives awards for customer service. Whenever USAA talks, the rest of the insurance industry should listen.) This matters. During this session, panelists polled the audience of high-level IT professionals about how they would use Watson. More than half indicated they’d use it better understand their customers.
Terry Jones, founder of Travelocity and founding chairman of Kayak.com, also announced WayBlazer, a Watson-powered travel service that blends data sources so users can do real-language, Siri-like queries and receive complex result. If I say “I want to use my miles to take a romantic vacation with my wife,” for example, the results will list all-inclusive vacation choices prioritized by my preferences that I could buy using my miles with one click. (I’ll tell you what: If Travelocity pulls that off, I’ll be a customer for life.) The demo worked surprisingly well, even providing warnings that might prompt users to choose another option else based on little known information.
Teach Your Machines Well, Their Bad Output Did Slowly Go By
As millennials and even younger customers some to market, they will reap the benefits of Watson and won’t necessarily need to become subject matter experts. Many of them won’t assure the accuracy of data going into the system but will likely rely on the highly compromised results that come out. Crap.
[ Commentary: Data Analytics Will Fail If Executives Ignore the Numbers ]
Actually, the takeaway should be that, done right, systems such as Watson could save your life or, at the very least, make a positive difference and assure the success of your firm. Done wrong – provided you don’t do QA on the data coming in and the decision coming out – you’re pretty much screwed. Not only do you need something like Watson, you absolutely need to make sure it’s implemented correctly.