Robotics Reference
In-Depth Information
bits with qubits we vastly enhance computational performance. This en-
hancement benefits from the fact that quantum computations can take
place simultaneously, thereby enabling quantum computers to perform
its computations in parallel. This parallelism creates an additional vast
increase in computing speed.
Quantum computing will allow us to process complicated informa-
tion faster. Its main applications will lie in complex computational tasks,
such as the encryption and decryption of confidential information, voice
recognition, image and shape recognition and other applications in arti-
ficial intelligence. Simpler computational tasks such as word processing
are unlikely to gain in performance with quantum computing.
Computer Memory
When considering advances in computer technology it is the speed of
the computers that attracts most attention, hence the great interest in
Moore's Law. But speed alone cannot be responsible for the exponential
growth in the capabilities of Artificial Intelligence systems in the decades
ahead. Computer memory sizes must also increase dramatically, partly
to allow a computer's processor to have more “workspace” in which to
perform its calculations and its manipulations of data, and partly to cater
for the ever increasing amounts of data needed by AI programming. Here
it is worthwhile repeating the wise words of Yorick Wilks, who has long
been one of the world's leaders in Natural Language Processing:
Artificial Intelligence is a little software and a lot of data. [6]
And a lot of data needs a lot of computer memory to store it.
In January 1979, when I started in the business of designing and pro-
gramming Chess computers, companies manufacturing consumer elec-
tronic products were paying approximately $10 per kilobyte
12
of “static”
RAM (Random Access Memory—the type of memory chip needed for
the processor's calculations), and $1 per kilobyte for ROM (Read Only
Memory—the type of memory chip used to store the program and its
data). At the time of writing (late 2004) the cost of static RAM is down
to around one cent per kilobyte (one-thousandth of the cost 25 years ago)
while the same type of ROM chips are now costing around $1 per three
12
1,000 bytes (or, to be more precise, 1,024 bytes, since computer memory sizes are normally
discussed in powers of 2). A byte is a
word
—a unit of computer memory, consisting of eight bits
(1s and 0s).
