Like many, I was glued to the TV (well, glued to my DVR’ed recording anyway) to see IBM’s “Watson” computer take on two of the world’s best human Jeopardy players. Watson handily beat the humans, prompting questions about whether this was the beginning of sentient machines that would eventually become our thankless butlers, or ruthless overlords depending on your vision of the future.
IBM’s development of Watson is certainly a good thing in terms of human achievement, but what Watson does should not be confused with an ability to “think.” Basically, what has been created with Watson is the ability to retrieve information based on a spoken question. While it is a far cry from sentience, it is still an amazing achievement in computing. When you think about the nuances of human language this becomes apparent. If you say “I slept with your sister”, I would likely be highly offended, but a computer might see nothing wrong with this statement, not knowing the connotations of “sleep” in this sentence. As another example, Watson could instantly tell you who said ‘Where’s the beef” in the 1980’s, but could not tell you which cheeseburger joint he prefers, or wax poetic about the subtle nuances between the Whopper and Big Mac.
What Watson gives us is, in effect, a really smart, really fast Google that responds to human language. This could be used in the altruistic examples IBM showed during the Jeopardy series (helping doctors around the world for free), or more likely replace a whole room full of call center agents. A system like Watson obviously isn’t free to build, deploy, and maintain, so I’d speculate we’ll see more of the latter, and be hearing a slightly-spooky robotic voice when we call tech support in the not-too-distant future.