Watson vs Siri – A natural language showdown.  The ultimate battle of the sexes

Having now had some time to use and digest the capabilities of Apples integration of Siri into it’s new iPhone 4S and previously having understood the capabilities of Watson from IBM I thought it was timely to look as what is becoming the new battlefield of voice recognition.

Mobile voice recognition has been around for a long time.  I recall using it back on Windows Mobile 5 many years ago now.  Some implementations required you to record the voice tag to the contact and then it simply matched the tag and applied the action.  Soon after the need for the voice tag was removed and the software looked up the contact or activity via speech to text translation.  This type of voice interaction (ie: Call Randall, open email etc.) is basically dumb voice matching and should not be compared to a natural language engine like Siri or Watson.

What is natural language

The first place to start is to watch this video of IBM’s Watson in action.

Quite amazing right?  Less than a year after Watsons public debut he now has his first job.   The first Watson deployment will be with WellPoint nurses who manage complex patient cases and review treatment requests from medical providers.   Imagine Watson listening in the background in every doctors consultation room throughout the world.  I feel a bit weezy, and crook (Aussie slang for sick) or I feel tight and gunky in the chest.  Watson needs to work out what the person is actually saying and learn from it.  What is the context of weezy, what does crook mean and how does gunky relate to chest and tight?  Simple for us mere humans but a real challenge for a computer and some software to actually ‘understand’.

So lets leap forward 10 years.  Watson is sitting in the corner and listening to 1000’s of conversations and doctors diagnosis of those conversations.  After the doctor diagnoses your condition he turns to Watson and says.  Watson do you agree?  He reply’s with based on the information provided I estimate with 89% accuracy that you are correct.   Rolling up all the info outbreaks of disease could be isolated and contained faster than ever.  This really is ground breaking technology.

Coming out swinging from the other corner is Apple’s Siri.  With its consumer focused deployment on millions of iPhones it is basically the opposite approach to IBM’s Watson.  Siri in a lot of ways is similar in architecture in that it sends all voice commands back to the mother ship for diagnosis and translation and then delivers the result back to the phone.  So with millions and millions of requests coming in each day can Siri learn faster that Watson?  Will it become the platform of choice?  Will consumerisation win the war.  Sell to the consumer and let them take it to work then let the enterprise work out how to support it vs Sell it to the enterprise approach of IBM.  Apples approach certainly seems to be working for them at present with iOS penetration into the enterprise at an all time high.  Can they caplitalise on this and sell server software with Siri to enterprise customers or will they keep the secret sauce to themselves to sell more Apple devices?

Interesting times ahead.