The fourth important idea of the information era, according to Kurzweil's introduction to Von Neumann's book, is to "endow computers with intelligence", i.e. to give computers the same thinking capacity as the human brain. This idea is centrally based on Turing's own paper "Computing Machinery and Intelligence". In his paper, Turing claims that instead of questioning if machines can "think" (which could lead to a complicated philosophical debate on the nature of thinking), we should instead consider on whether or not a machine can imitate the human thinker. This is then further elaborated in von Neumann's book, which Kurzweil claims to be exploiting on the computer's potential capacity to accurately emulate the brain’s processing. Turing also makes a brief claim that our conceptions and criticisms of the limits of the computer is just a parallel of the limited storage capacity of decades past, when the typical storage of today were unthinkable. It is still a big debate in our days, however, if computers are (or will) be able to perceive emotions and express them when provided with increasing storage capacity.
On the other hand, Von Neumann underlines that the speed of neural processing is extremely slow which contradicts the race for higher computing power by the use of supercomputers. In order to underline the parallel nature of neural processing, he contrasts the individual neuron latency of the computing in the human brain to its high bandwidth in the throughput of 10^10 neurons simultaneously. By this means, he demonstrate an actual flaw in his computer architecture model which is not imitating the parallel nature of computing in the humans. Nevertheless, he estimates that the supercomputers will be able to simulate the human brain with 10^16 operations per second with the supercomputers in 2020.
No comments:
Post a Comment