I usually write about education in this column. But, yesterday, the New York Times ran a front page article on Artificial Intelligence. They ran it because there is an upcoming competition between an IBM computer and the champions of the Jeopardy TV show. It is being billed as a man against machine competition to see if people are smarter than computers or vice versa.
Whenever there is nonsense to print these days, the Times seems to be right on it. The time claims that:
“machines (have begun) to “understand” human language. Rapid progress in natural language processing is beginning to lead to a new wave of automation that promises to transform areas of the economy that have until now been untouched by technological change.”
Long before I worked on education I was a leader in the field of Artificial Intelligence. My specialty was Natural Language Processing. I worried about how computers could possibly understand language in the same way that humans understand language. I came to the conclusion that while this was a daunting task, it was probably not an impossible one. But, in order to make computers understand language they would need dynamic memories and they would need to be able to learn (because what you hear and read changes what you know). They would also need goals (because we understand in terms of what we care about) and plans, because we learn in order to help us do something better. I began to work on learning and memory, and understanding how planning works. And, while there has been much progress in AI in those areas, we are still far from having very intelligent machines that can do such things very well.
Not according to the New York Times, of course. There headline was
SMARTER THAN YOU THINK
A Fight to Win the Future: Computers vs. Humans
Gee, will computers suddenly take over? I have been asked this question by every reporter and TV person who ever interviewed me about AI. The nonsense behind this question is too long to discuss her. But here is what the Times said:
“Machines will increasingly be able to pick apart jargon, nuance and even riddles. In attacking the problem of the ambiguity of human language, computer science is now closing in on what researchers refer to as the “Paris Hilton problem” — the ability, for example, to determine whether a query is being made by someone who is trying to reserve a hotel in France, or simply to pass time surfing the Internet.”
All this because a computer will try to play Jeopardy.
Computers have been getting by for decades now on key word search. Google has made key word search an art form. The “Paris Hilton” problem is not a problem for people however. In spoken English, the hotel is pronounced with an emphasis on Paris (as opposed to London.) But, people don’t really need that spoken cue so much because context tells you what is being talked about. We see or read about Paris Hilton. We make a reservation at the Paris Hilton. “The food is bad at the Paris Hilton” is not a confusing sentence. It is only confusing to a computer that doesn't know what you are talking about and processes only key words. In other words, the Times is discussing ideas about how to use statistics to make a best guess about what the words might mean. And then, seeing that a program might be good at this, the Times then predicts the takeover of mankind by smart computers.
The New York Times used to be a great newspaper. I have subscribed for over 40 years. But these days much of what they have to say is nonsense. When Bryant Gumbel asked me on the Today Show, many years ago, whether computers would soon take over, I attributed his question to the need for sensational junk on morning TV. The MacNeill Lehrer Report on PBS asked sensible questions. Redes in Spain asked sensible questions. But, alas the Times doesn’t care that the average reader is going to draw conclusions about a computer’s ability to understand that simply aren’t true. And I don’t think they give a damn.