International Business Machines researchers spent four years developing Watson, the computer smart enough to beat the champions of the quiz show "Jeopardy!" Now they're trying to figure out how to get those capabilities into the phone in your pocket.
Bernie Meyerson, IBM's vice president of innovation, envisions a voice-activated Watson that answers questions, like a supercharged version of Apple's Siri personal assistant. A farmer could stand in a field and ask his phone, "When should I plant my corn?" He would get a reply in seconds, based on location data, historical trends and scientific studies.
Finding additional uses for Watson is part of IBM's plan to tap new markets and boost revenue from business analytics to $16 billion by 2015. After mastering history and pop culture for its "Jeopardy!" appearance, the system is crunching financial information for Citigroup and cancer data for WellPoint. The next version, dubbed Watson 2.0, would be energy- efficient enough to work on smartphones and tablets.
"The power it takes to make Watson work is dropping down like a stone," Meyerson said in an interview. "One day, you will have ready access to an incredible engine with a world knowledge base."
IBM expects to generate billions in sales by putting Watson to work in finance, health care, telecommunications and other areas. The computer, which 15 million people saw beat former "Jeopardy!" champions Ken Jennings and Brad Rutter, is the company's most high-profile product since it sold its personal- computer unit to Lenovo Group seven years ago.
The challenge for IBM is overcoming the technical obstacles to making Watson a handheld product, and figuring out how to price and deliver it. Watson's nerve center is 10 racks of IBM Power750 servers running in Yorktown Heights, N.Y., that have the same processing power as 6,000 desktop computers. Even though most of the computations occur at the data center, a Watson smartphone application would still consume too much power for it to be practical today.
Another hurdle: It takes awhile for Watson to do the "machine learning" necessary to become a reliable assistant in an area. Watson's deal with WellPoint was announced in September of last year, and the system won't master the field of oncology until at least late 2013.
Researchers also need to add voice and image recognition to the service so that it can respond to real-world input, said Katharine Frase, vice president of industry research at Armonk, N.Y.-based IBM.
"In 2.0, we hope to give him more senses," Frase said. "A guy could say into his phone, 'Here's where I am and here's what I see,' lifting it up to take in images of the environment."
IBM's path to the mobile-assistant market contrasts with Apple's. For one, IBM is focused on corporate customers, while Apple is targeting anyone who buys its phones.
Apple made Siri the focus of its marketing of the iPhone 4S, which debuted last year. The software is touted as a personal assistant that can answer a wide range of spoken questions -- "Do I need an umbrella tomorrow?" -- and put appointments in a calendar.
Siri has become a defining characteristic of the iPhone, though it's also drawn complaints. In a June survey by Minneapolis-based Piper Jaffray & Co., Siri was found to resolve requests correctly less than 70 percent of the time.
Trudy Muller, a spokeswoman for Cupertino, Calif.-based Apple, said customers are happy with Siri and the company will further improve the software in the next version of the iOS operating system.
"Siri is one of the most popular features of iPhone 4S," she said. "With new features coming this fall in iOS 6, Siri will help you get even more done with just your voice."
With Watson, IBM aims to tackle more complex questions. The program will be able to understand oncology well enough to advise doctors on diagnosis and prescriptions, said Martin Kohn, IBM's chief medical scientist. One iPad application for Watson -- a health-care program developed with a Columbia University professor -- is being used to demonstrate its medical capabilities for prospective IBM customers.
As the technology is improved, a mobile Watson could become an extension of services that IBM already offers to business customers such as WellPoint, the second-biggest U.S. health insurer. The move fits into a broader push to promote analytics software, which helps customers diagnose problems and spot patterns in everything from infant mortality to South American floods. IBM agreed to buy Kenexa Corp. for about $1.3 billion yesterday as part of the effort.
In one potential scenario, a patient accesses Watson via a mobile device to explain symptoms in natural language. The person then gets several recommendations for what might be happening, listed in order of the computer's confidence. The patient and a doctor both have access to Watson, which would securely access the patient's medical records, letting it make adjustments to an answer depending on factors like pregnancy or diabetes, Kohn said.
For now, the features are being developed for care providers, not patients themselves, and the program is still in the testing stage.
"There's going to come a point where you don't need any intermediary," Meyerson said.
To "teach" Watson a subject such as oncology, researchers feed the machine with answers to relevant questions. They then ask it to answer similar queries by analyzing documents, websites and books at 66 million pages a second. After enough times telling the machine which answers are accurate, Watson can develop expertise to assist a doctor.
"It's going to recommend to you the answer and show you the evidence," Frase said.
Adding voice recognition and other senses to Watson might be easier than adding knowledge because IBM already makes tools that understand images and natural language, Kohn said. An IBM project for the U.S. military, for example, translated English into local dialects of Arabic.
"Watson itself will not interpret images, but IBM has technology that does interpret images and numerical information," Kohn said. "Watson 2.0 is all of these different tools, working together."