Communications of the ACM
Why We’re a Long Way From Computers That Really Work Like the Human Brain
No computer today is sufficiently powerful to run a program that simulates the human brain.
Credit: Fergal Crehan
Despite recent advances in computers designed to simulate the human brain, significant work remains to be done before this technology achieves its goal.
IBM, for example, recently announced a computing architecture based on the human brain, as part of its SyNAPSE project. The new architecture is a programming language designed for chips that reduce the gap between memory and processing power to make them faster, more compact, and more power-efficient.
Meanwhile, German scientists led by computational neurophysicist Markus Diesmann are working on the Human Brain Project.
No computer currently is powerful enough to run a program that simulates the brain, due in part to the brain’s interconnected nature. In computing, processors serve as the brain’s neurons, and memory functions as the synapses where neurons meet and transmit information to each other. Commercial chips can hold billions of transistors, just as the brain holds roughly 100 billion neurons. However, transistors typically have only three connections, whereas a neuron can have up to 10,000 points of connection; and a brain has about 100 trillion synapses.
"There's no chip technology which can represent this enormous amount of wires," Diesmann says. He estimates that a fully working simulation of the brain will not emerge for another 10 to 20 years.
From Quartz
View Full Article
Abstracts Copyright © 2013 Information Inc., Bethesda, Maryland, USA
No entries found