updated 04:32 pm EDT, Wed August 29, 2012
Efficiency improvements to make Siri-competitor possible
IBM is looking to bring the technology that powers Watson to smartphones. The technology giant says that advances in both battery power and computing efficiency make it possible to bring a new version of the smart computer, dubbed Watson 2.0, to mobile handsets within the coming years. The new version will bring the voice recognition capabilities the system is already known for, but IBM is also looking to add more senses to the technology, allowing Watson to recognize images as well.
While Apple's Siri and other digital assistants have mainly targeted the consumer sector, Watson will reportedly be aimed at corporate clients. IBM is developing the technology to be able to handle complex questions across health care, finance, and telecommunications, among other areas. Watson has already been put to work in some medical fields, assisting oncologists in diagnosis and prescribing, though it will likely not achieve mastery of the field for another few years.
Making Watson a capable digital assistant will require more advances in software efficiency. Most of the processing load for Watson is handled on 10 racks of IBM Power750 servers located in Yorktown Heights, NY. That equates to about 6,000 desktop computers. Still, a smartphone version of Watson would likely consume too much power to make it a practical option for today's mobile devices. Additional obstacles include the time it takes for Watson to do the requisite machine learning to become sufficiently versed on a subject.
Watson 2.0 will reportedly integrate a number of other technologies IBM has developed over the past several years. The company has been working on image recognition and natural language processing for the US military, and these technologies are said to be targets for the next generation of Watson. The commercialization of Watson has been previously targeted as a growth initiative for IBM, which plans to boost revenues from business analytics to $16 billion by 2015. [via SFGate]