122
PHYSICS OF THE IMPOSSIBLE
stat that monitors the temperature in a room to the self-aware organisms that we are today. Animals may be conscious, but they do not possess the level of consciousness of a human being. One should try,
therefore, to categorize all the various types and levels of consciousness
rather than debate philosophical questions about the meaning of consciousness. Robots may eventually attain a "silicon consciousness." Robots, in fact, may one day embody an architecture for thinking and
processing information that is different from ours. In the future, advanced robots might blur the difference between syntax and semantics,
so that their responses will be indistinguishable from the responses of
a human. If so, the question of whether they really "understand" the
question will be largely irrelevant. A robot that has perfect mastery of
syntax, for all practical purposes, understands what is being said. In
other words, a perfect mastery of syntax is understanding.)
COULD ROBOTS BE DANGEROUS?
Because of Moore's law, which states that computer power doubles
every eighteen months, it is conceivable that within a few decades robots will be created that have the intelligence, say, of a dog or a cat. But
by 2020 Moore's law may well collapse and the age of silicon could
come to an end. For the past fifty years or so the astounding growth in
computer power has been fueled by the ability to create tiny silicon
transistors, tens of millions of which can easily fit on your fingernail.
Beams of ultraviolet radiation are used to etch microscopic transistors
onto wafers made of silicon. But this process cannot last forever. Eventually, these transistors could become so small that they reach the size
of molecules, and the process will break down. Silicon Valley could become a Rust Belt after 2020, when the age of silicon finally comes to
an end.
The Pentium chip in your laptop computer has a layer about
twenty atoms across. By 2020 that Pentium chip might consist of a
layer only five atoms across. At that point the Heisenberg uncertainty
Précédent

- 146/353

Suivant