Manufacturing Innovation Lessons From IBM's Watson
Posted by Randall Cronk on Thu, Feb 24, 2011 @ 08:10 AM
When IBM’s Watson computer played
Jeopardy! against two of the game’s greatest champions, it was hard to remember you were watching a machine. (IBM has been promoting Watson’s
Jeopardy! Challenge on its homepage for weeks.) Yes, it’s a computer so it can remember every fact it was ever “taught” — covering the equivalent of millions of pages of documents. It also doesn’t get nervous in front of the cameras or worry about losing all its winnings if it wagers too much on the final
Jeopardy! category. Watson is also lightening fast pushing its buzzer. And, again, you would expect that of a machine.
But being faster, or cool under pressure, or able to remember vast numbers of facts isn’t that impressive or distinctive. Computers are like that, it’s true, but then so are a lot of Jeopardy! players. What was eerie about watching Watson take apart the opposition was the game itself — the fact that this is Jeopardy!
We’ve seen computer-versus-human matchups before. Remember when IBM’s Deep Blue beat chess grandmaster Garry Kasparov? But chess is different from Jeopardy! Chess is complex, yes, but at the end of the day it boils down to finite number of moves and countermoves in response to previous moves and countermoves. The number of relevant combinations for winning is not that high by standards of a modern computer. More importantly, they are just moves.
But Jeopardy! is about language as a human system to express and understand knowledge, something that can’t be programmed with rules. To illustrate the problem, IBM recalls an old Groucho Marx joke in its videos: “Last night I shot an elephant in my pajamas. How it got in my pajamas I’ll never know.” Humans get the fact that the elephant isn’t wearing pajamas. Watson does too.
True Manufacturing Innovation
For IBM, of course, Watson is no joke. It is the ultimate expert system. It doesn’t just have total recall and the ability to correctly interpret English syntax based on the content of a sentence. It also weighs the relevance and reliability of information to make an accurate wager as to whether its answers are correct.
In other words, Watson is to answers what Google is to search. Imagine a for-production-engineers-only version of Jeopardy! deployed as a web service. Or a repair-technician-only version. An OEM could simply upload all its maintenance and parts information to the cloud once to enable each tech at every customer site to answer virtually any maintenance question instantly. Answers could be graded by levels of confidence, and links to underlying information made available.
Think what Watson could mean for consumers looking for technical help with their PCs or home media centers. Customer service would be transformed. Product companies looking for a way to invigorate their service offering may have just found a new friend — Watson.
Photo Credit:charliecurve