Monday, February 21, 2011

"Watson, come here! I need you" - These were reportedly the first words spoken by Alexander Graham Bell to his assistant Thomas Watson on his "electrical speech machine", which would later be called the telephone. With this one sentence, Bell broke a seemingly insurmountable barrier of enabling live conversations over long distances. Until that time in 1876, the telegraph was the only thing that allowed long distance communications, that too only one way at a time. His new technology allowed the same infrastructure of installed telegraph wires to send live conversations. To the 19th century America, it probably sounded like magic. In a sense, he had broken the figurative "sound barrier".

Fast forward to last week. My entire family was watching the show "Jeopardy". Everyone was glued to the set for three days, since it was the new "Watson", a computer built by IBM, going up against two of the best human Jeopardy players ever. At the end of the first day, the computer was tied with one of the players. By the end of the second day, it was no contest. The computer pulled away rapidly and finished first. By the end of the two round tournament, Watson had won more money than the two remaining contestants combined. One of the other contestants, Ken Jennings, wrote a humorous remark in his Final Jeopardy answer welcoming the "new overlords", i.e., the computers.

So, what is so special about Watson? To get the answer, needs to go back to some history of computing. During my years at Intel, I had the opportunity to discuss speech recognition with leading edge researchers at IBM's TJ Watson Research Center in New York. They were heady times as Moore's Law routinely brought twice the computing power into the field every two years. By 2008, there were more than a billion transistors produced for every man, woman and child on this earth. And yet, the "sound barrier", so to speak, was not broken. What I mean is the ability of computers to understand natural language. The kind that movie goers watched in "2001 A Space Odyssey" as the astronaut commands computer HAL to "open the pod bay door, HAL", HAL rebuts him by saying "I'm sorry Dave, I'm afraid I can't do that". To the human brain, this skill is learned by age two or three. And yet, training a computer to understand natural language has eluded scientists for a long time, until now. Watson's accomplishments are even more significant, in that the computer had to understand the quirks of the game, unusual contexts, and go up against two of the best players ever to play on TV. On the minus side, Watson only had to decipher the text, not the spoken words. But I am guessing integrating it is not that far off either, since many commercial programs like Dragon Naturally Speaking already convert spoken words to text.

What does this mean for the future? For now, at least, Watson is a supercomputer in IBM research lab that fills an entire room, and was programmed by some leading research scientists. However, the same can be said for the Eniac, the first computer ever built, circa 1950. The four function pocket calculators that had more processing power than the Eniac followed less than two decades later, using the newly invented transistor technology. Since then, the pace of innovation in computer hardware has accelerated to the point where we may have Watson like technology in our palms less than a decade from now. Elements of such technology are already in use and ubiquitous today. These can already be put to good use to help our students learn new skills and concepts much faster than traditional methods, especially in fields like mathematics.

These are exciting times to be consumers of technology, and the future looks even better.

No comments: