Australian and Japanese scientists have discovered that an artificial network of nanowires may physically function at its peak at the ‘edge of chaos’, much like the human brain.
The team, led by Joel Hochstetter of the University of Sydney, ran computer simulations to test how a random nanowire network, a type of artificial intelligence, best performs tasks. They found that the wires acted almost like neurons.
The information processing in the nanowire network was physical and required minimal direction once stimulated, much like the brain, but problem-solved better with the right level of stimulation.
When the signal stimulating the network was too low, there was too much order and predictability for it to produce complex outputs, but when there was too much stimulation, the output was chaotic and useless for problem solving.
“We found that if you push the signal too slowly, the network just does the same thing over and over without learning and developing. If we pushed it too hard and fast, the network becomes erratic and unpredictable,” says Hochstetter.
Instead, the peak performance was achieved when the signal fell just short of this chaotic stimulation, suggesting that, like the brain, the Goldilocks of performance was at the edge of chaos.
“Some theories in neuroscience suggest the human mind could operate at this edge of chaos, or what is called the critical state,” says Professor Zdenka Kuncic from the University of Sydney, who supervised Hochstetter. “Some neuroscientists think it is in this state where we achieve maximal brain performance.
“What’s so exciting about this result is that it suggests that these types of nanowire networks can be tuned into regimes with diverse, brain-like collective dynamics, which can be leveraged to optimise information processing.”
Unlike normal computers, where memory (RAM) and operations (CPU) are separate, the AI nanowire network had these two processors as a single system because of junctions between the wires. These physical junctions acted like switches that depended on historic response to electrical signals, switching on and off to allow current to flow between the wires.
“These junctions act like computer transistors but with the additional property of remembering that signals have travelled that pathway before,” says Hochstetter. “As such, they are called ‘memristors’. This creates a memory network within the random system of nanowires.
“Where the wires overlap, they form an electrochemical junction, like the synapses between neurons.
“We found that electrical signals put through this network automatically find the best route for transmitting information. And this architecture allows the network to ‘remember’ previous pathways through the system.”
The study was published in Nature Communications.