Machines – Every January, technology writers reach for the same tired prediction: this is the year artificial intelligence finally arrives. We have been making that promise, in one form or another, since a small group of researchers gathered at Dartmouth in the summer of 1956 and confidently declared that a serious crack at machine intelligence might take a couple of months. Fifty-five years later, the couple of months has stretched into a couple of lifetimes, and the machines still cannot reliably tell a photograph of a cat from a photograph of a dog.
And yet, this year, the old prediction feels less foolish than usual.
Happy New Year. It should be an interesting one.
Consider what is about to happen in February. IBM has built a computer called Watson, and in a few weeks it will step onto the Jeopardy! stage to play against Ken Jennings and Brad Rutter, the two greatest human champions the show has ever produced. This is not chess. When Deep Blue beat Garry Kasparov back in 1997, it did so by brute calculation — searching millions of board positions per second within a game whose rules are fixed and finite. Jeopardy! is a different animal entirely. The clues are full of puns, riddles, half-buried references, and the kind of slippery wordplay that human beings absorb without thinking and computers choke on completely. If Watson can untangle that, even some of the time, it will mark something genuinely new: a machine wrestling with the messiness of ordinary human language and occasionally winning.
The interesting part is not whether Watson wins the million-dollar prize. The interesting part is what IBM intends to do with it afterward. The company is already talking about pointing the same technology at medicine — feeding it the mountain of journal articles and case histories no human doctor could ever read in a lifetime, and asking it to suggest diagnoses. Whether that pans out is anyone’s guess. But the ambition tells you where the wind is blowing.
What has changed, really, is not some sudden breakthrough in machine cleverness. It is data, and it is scale. We now live in a world that produces an almost unimaginable quantity of digital information every single day — photos uploaded to Facebook, videos to YouTube, billions of searches typed into Google. And we have, finally, cheap enough computing power to do something useful with all of it. The fashionable phrase this year is “the cloud,” which mostly means renting somebody else’s enormous warehouse of servers instead of buying your own. The practical upshot is that a clever startup with no money can now train its software on quantities of data that would have required a national laboratory a decade ago.
Google is the company that understood this earliest and best. Its spam filters, its translation tools, its eerily accurate search suggestions — none of these work because an engineer sat down and taught the computer the rules of grammar or the definition of junk mail. They work because the system was shown staggering numbers of examples and allowed to find the patterns itself. This is the quiet revolution: we have largely stopped trying to program intelligence directly, and started trying to grow it from data. The results are often crude, but they are improving at a pace that ought to give us pause.
I should be careful not to oversell any of this. The truth is that today’s “artificial intelligence” is narrow to the point of idiocy. A program that can recommend a book you might like cannot also recognize a face, and a program that recognizes faces cannot drive a car, and none of them have the faintest idea what they are doing or why. They possess no common sense, no understanding, nothing we would recognize as a mind. The grand dream of the Dartmouth pioneers — a machine that thinks the way a person thinks — remains exactly as distant as it was when they first dreamed it. Anyone who tells you otherwise is selling something.
We have been burned before, too. The 1980s saw a wave of enthusiasm for “expert systems” that was supposed to revolutionize everything and instead collapsed into what veterans of the field still call the AI winter — years of broken promises, evaporated funding, and embarrassed silence. The smart money has learned to be skeptical of grand pronouncements.
But something does feel different now, and I think it is this: the useful, unglamorous version of artificial intelligence has quietly stopped being science fiction and become infrastructure. It is in your phone’s predictive text and your email’s spam folder and the route your GPS chooses through traffic. It does not announce itself. It does not pass for human. It simply works, a little better each year, until one day you realize you cannot remember how you managed without it.
So no, the machines will not become genuinely intelligent in 2011. We will not build a mind this year, or next year, or — I suspect — for a very long time to come. But we are learning, finally, to make our machines useful in ways that look a little like thinking. And if Watson holds its own against the best of us next month, it will be worth remembering that the most important revolutions rarely arrive with trumpets. They arrive disguised as conveniences.
On the other hand, today’s technologies are marriage with new and glittery new hardware to make consumer life easy the connection between the terminal equipment and the transceiver can be realized with a serial cable (e.g., USB), a Bluetooth link, an infrared link, etc. Common AT commands based on some science include AT+CMGS (send a message), AT+CMSS (send a message from storage), AT+CMGL (list messages) and AT+CMGR (read the message). However, not all modern devices support receiving of messages if the message storage (for instance the device’s internal memory) is not accessible using AT commands.
====================== About the Author =================================
Read about Author at : About Me
Thank you all, for spending your time reading this post. Please share your feedback / comments / critics / agreements or disagreement. Remark for more details about posts, subjects and relevance please read the disclaimer.
FacebookPage Twitter ContactMe LinkedinPage =========================================================================
