"The power it takes to make Watson work is dropping like a stone," IBM’s vice president of innovation Bernie Meyerson told Bloomberg. "One day, you will have ready access to an incredible engine with a world knowledge base."
Image Credit: Seth Wenig, AP Photo
Could "Watson 2.0" be the Cadillac of voice-activated virtual assistants? For now, the Siri-Watson comparison is more like comparing a Prius to a Boeing 747: IBM’s current version runs on ten racks of Power750 servers, with the processing muscle of 6,000 desktop computers. An app that just relayed questions to and from IBM’s data center would require too much power for any current smartphone to run it at a practical speed.
And while IBM envisions Watson 2.0 as being efficient enough to run on current mobile hardware, the program has lots of "machine learning" to do if it’s going do more than search Wikipedia or IMDb. Siri’s abilities to trawl Yelp for restaurants and integrate with your calendar and reminders took years of programming and development. So while Watson is already way better at decoding natural language, the system has to be "taught" how to handle any new utility functions.
For now, Watson is primarily being used and developed for business analytics. But IBM’s ongoing commitment to portability could lead the industry-wide charge toward better voice-recognition. Siri better bone up on trivia if she wants to stay in the game.
Source : feeds[dot]ign[dot]com