-
[ Pobierz całość w formacie PDF ]
As the computing cloud grows, as it becomes ubiquitous, we will feed ever more
intelligence into it. Using global positioning satellites and tiny radio transmitters, it will track our
movements through the physical world as meticulously as it today tracks our clicks through the
virtual world. And as the types of commercial and social transactions performed through the
Internet proliferate, many more kinds of data will be collected, stored, analyzed, and made
available to software programs. The World Wide Computer will become immeasurably smarter.
The transfer of our intelligence into the machine will happen, in other words, whether or not we
allow chips or sockets to be embedded in our skulls.
Computer scientists are now in the process of creating a new language for the Internet
that promises to make it a far more sophisticated medium for expressing and exchanging
intelligence. In creating Web pages today, programmers have limited options for using codes, or
tags, to describe text, images, and other content. The Web s traditional hypertext markup
language, or HTML, concentrates on simple formatting commands on instructing, for instance,
a Web browser to put a line of text into italics or to center it on a page. The new language will
allow programmers to go much further. They ll be able to use tags to describe the meaning of
objects like words and pictures as well as the associations between different objects. A person s
name, for instance, could carry with it information about the person s address and job, likes and
dislikes, and relationships to other people. A product s name could have tags describing its price,
availability, manufacturer, and compatibility with other products.
This new language, software engineers believe, will pave the way for much more
intelligent conversations between computers on the Internet. It will turn the Web of
information into a Web of meaning a Semantic Web, as it s usually called. HTML s
inventor, Tim Berners-Lee, is also spearheading the development of its replacement. In a speech
before the 2006 International World Wide Web Conference in Scotland, he said that the Web is
only going to get more revolutionary and that twenty years from now, we ll look back and say
this was the embryonic period. He foresees a day when the mechanisms of trade, bureaucracy
and our daily lives will be handled by machines talking to machines.
At the University of Washington s Turing Center, a leading artificial intelligence
laboratory, researchers have already succeeded in creating a software program that can, at a very
basic level, read sentences on Web pages and extract meaning from them without requiring
any tags from programmers. The software, called TextRunner, scans sentences and identifies the
relationships between words or phrases. In reading the sentence Thoreau wrote Walden after
leaving his cabin in the woods, for instance, TextRunner would recognize that the verb wrote
describes a relationship between Thoreau and Walden. As it scans more pages and sees
hundreds or thousands of similar constructions, it would be able to hypothesize that Thoreau is a
writer and Walden is a book. Because TextRunner is able to read at an extraordinary rate in one
test, it extracted a billion textual relationships from 90 million Web pages it can learn quickly.
Its developers see it as a promising prototype of machine reading, which they define as the
automatic, unsupervised understanding of text by computers.
Scientists are also teaching machines how to see. Google has been working with
researchers at the University of California at San Diego to perfect a system for training
computers to interpret photographs and other images. The system combines textual tags
describing an image s contents with a statistical analysis of the image. A computer is first trained
to recognize an object a tree, say by being shown many images containing the object that
have been tagged with the description tree by people. The computer learns to make an
association between the tag and a mathematical analysis of the shapes appearing in the images. It
learns, in effect, to spot a tree, regardless of where the tree happens to appear in a given picture.
Having been seeded with the human intelligence, the computer can then begin to interpret
images on its own, supplying its own tags with ever increasing accuracy. Eventually, it becomes
so adept at seeing that it can dispense with the trainers altogether. It thinks for itself.
In 1945, the Princeton physicist John von Neumann sketched out the first plan for
building an electronic computer that could store in its memory the instructions for its use. His
plan became the blueprint for all modern digital computers. The immediate application of von
Neumann s revolutionary machine was military designing nuclear bombs and other
weapons but the scientist knew from the start that he had created a general purpose technology,
one that would come to be used in ways that could not be foretold. I am sure that the projected
device, or rather the species of devices of which it is to be the first representative, is so radically
new that many of its uses will become clear only after it has been put into operation, he wrote to
Lewis Strauss, the future chairman of the Atomic Energy Commission, on October 24, 1945.
Uses which are likely to be the most important are by definition those which we do not
recognize at present because they are farthest removed from our present sphere.
We are today at a similar point in the history of the World Wide Computer. We have built
it and are beginning to program it, but we are a long way from knowing all the ways it will come
to be used. We can anticipate, however, that unlike von Neumann s machine, the World Wide
Computer will not just follow our instructions. It will learn from us and, eventually, it will write
its own instructions.
GEORGE D YSON, A historian of technology and the son of another renowned [ Pobierz całość w formacie PDF ] - zanotowane.pl
- doc.pisz.pl
- pdf.pisz.pl
- zambezia2013.opx.pl