Engineers have developed ‘neuromorphic’ chips that can imitate some properties of that are similar to the human brain.
The human brain, although organic in nature, is possibly the most complex and most efficient ‘computer’ known to man. For quite some time, scientists and engineers have strived to re-create the most pivotal part of being human through technologies and materials that are currently available. Still, no artificial brain has come close to achieving what a real human brain can do.
Researchers from the University of Zurich and ETH Zurich and colleagues from the EU and US, however, have taken us a step closer to achieving the goal of constructing an artificial brain that’s just as beautiful as a real brain.
Using what they call ‘neuromorphic chips’, the researchers were able to create a complex electronic system that mimics some cognitive features of the brain. The team has shown that they can construct networks using artificial neurons (i.e. neuromorphic) so as to achieve the ‘human-like’ response by an AI system. Through the combination of the artificial neurons, they were able to produce processing modules that was said to be equivalent to that of ‘finite-state machines’.
The team demonstrated that their neuromorphic system can carry out complex motor tasks in real time, while exhibiting brain-like cognitive functions like drawing from short-term memory and context-dependent decision-making.
“The network connectivity closely resemble structures that are also found in mammalian brains,” Giacomo Indiveri, a professor at the Institute of Neuroinformatics, added. “Our Goal is to emulate properties of biological neurons and synapses directly on microchips.”
As a part of their goal, the team also hopes that their work in the neuromorphic system can push the boundaries AI by producing electronic circuits that are comparable to real brains in regards to size, speed, and energy consumption.