Machine Learning & Artificial Intelligence
As smartphones get smarter and speech recognition apps like SIRI and Google Now learn more about you personally and start acting on your behalf, offering information or suggestions before you even think to ask, what will that learned knowledge be worth? What if the knowledge is about you personally — your health needs and medications, your personal traits and preferences and habits, what makes you happy and makes you feel good, or what makes you money? Will AI developers be able to build new barriers-to-entry and gain a significant competitive advantage by treating collected knowledge as proprietary, making it hard to justify a shift to competing products?
In a previous article, I wrote about IBM’s Watson, a supercomputer that played the game of Jeopardy and won against the world’s best and brightest. It was taught to understand naturally spoken words and nuances of the English language, and it had the ability to read and analyze the equivalent of 300 million books in less than 3 seconds to find relevant information and determine the most likely answers based on probabilities (educated guesses). Now Watson is being applied to medical diagnostics to help physicians, but what if you personally had that sort of capability in the palm of your hand? Well, Apple and Google are moving in that direction. I was intrigued by two recent articles.
1. Google’s New Director Of Engineering, Ray Kurzweil, Is Building Your ‘Cybernetic Friend’ appeared in TechCrunch, describing AI software that knows users better than they know themselves. Google’s vision is to combine the personalized recommendations of a friend (who may know you better than you know yourself) with the sum of all human knowledge, creating a sort of super best friend.
This friend of yours, this cybernetic friend, may know that you’re struggling with certain health issues or business strategies. And, It may then canvass all the new information that comes out in the world every minute and bring things to your attention without you even asking about them.
Can’t live without it? What would it be worth if that information, or those suggestions, helped you make an important business decision, close a big account, and get a raise, and did this consistently all the time? What if, by monitoring your activity, diet, physiology, and personal traits, by working with Watson and medical practitioners, and by knowing your DNA, the system helped personalize the best new course of action to keep you healthy and achieve your fitness, financial and personal goals? What would that be worth? And how likely would you be to try out a competing system that doesn’t have the benefit of being uour super best friend for years?
2. SIRI RISING: The Inside Story Of SIRI’s Origins — And Why She Could Overshadow The iPhone describes how SIRI got it’s roots from a well funded military that brought together our nation’s top AI experts. The project showed that machines could learn in real time through what they sense and experience, as human beings do, rather than being “programmed” with a set of machine-readable rules. Apple bought the technology and significantly scaled back its functions to fit iPhone capabilities, but the original vision still lives, as described in the two videos below.
DARPA’s CALO project gave birth to SIRI
CALO was part of the PAL (Personal Assistant that Learns) program, funded by the Defense Department’s investment arm, DARPA.
CALO brings back memories.
While still working at IBM in the late 1970’s I sold a natural language, artificial intelligence query product called INTELLECT. This mainframe-based product had a build-in understanding of the English language at least 2-3 years before the PC. It didn’t take off in the market, however, because it was designed for senior executives, who lacked the required typing skills. I knew then that adding speech recognition would turn this sort of product into a success, as would pushing it down-market to consumers.
The simulated speech in the video sounds eerily similar to voices in IBM videos that were produced 30 years ago. Today the computing power of the $3.5 million IBM System/370 model 158-3 mainframe that I worked on back then is embedded in my Philips Sonicare toothbrush. Even more power is in the iPhone, making SIRI possible. But just imagine combining SIRI and today’s Watson, or what we’ll eventually see as we follow the trend curve of Moore’s Law.
As Google’s Ray Kurzweil predicts, sometime this year (2013), a supercomputer will exceed the computational power of the human brain. By 2023, a $1,000 computer will do that, and by 2037 a $0.01 computer will. By 2049, a $1,000 computer will exceed the computational power of the human race, and by 2059 a $0.01 computer will. (from my slides in Dr. Oz on Technology in Medicine.)